From 9b8836f950adfecfa84b797b4e6a4642d01510dd Mon Sep 17 00:00:00 2001
From: Carson <cpsievert1@gmail.com>
Date: Tue, 8 Sep 2026 11:09:01 -0500
Subject: [PATCH 1/7] fix(posit): switch API flavor when chat.model crosses
 model families

ChatPosit() picks its provider class from the model name (Claude models
get the Anthropic flavor, everything else the OpenAI-compatible one),
but the Chat.model setter only updated the model name, so switching to
a model from the other family sent requests in the wrong wire format
to the wrong gateway endpoint.

Adds a Provider.set_model() hook that the Chat.model setter consults;
the default returns the provider unchanged. The two Posit providers
override it to swap in their sibling class on a family switch,
carrying over credentials, the gateway base URL, and the cache
setting (stored inertly on PositOpenAIProvider so it survives a round
trip). Mirrors tidyverse/ellmer#1139.

Also drops unsigned thinking blocks when replaying history to
Anthropic-backed providers: reasoning emitted by a non-Claude model
has no signature, and the API rejects it with "Invalid signature in
thinking block". Family switching makes this reachable; Claude's own
(signed) thinking is unaffected.

Verified end-to-end against the Posit gateway with family switches in
both directions, including a conversation containing GLM thinking
content replayed to Claude.
---
 CHANGELOG.md                     |  2 +
 chatlas/_chat.py                 |  6 ++-
 chatlas/_provider.py             | 18 ++++++++
 chatlas/_provider_anthropic.py   | 16 ++++++-
 chatlas/_provider_posit.py       | 31 +++++++++++++
 tests/test_provider_anthropic.py | 38 ++++++++++++++++
 tests/test_provider_posit.py     | 75 ++++++++++++++++++++++++++++++++
 7 files changed, 182 insertions(+), 4 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 27c82e06..20139caf 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -13,6 +13,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Bug fixes
 
 * `ChatBedrock()` (with the default `api="converse"`) no longer sends assistant turns with an empty `content` array, which Converse rejects. This happens when a response carries no content blocks, for example when a guardrail intervenes before the model produces any. A `"[empty string]"` placeholder is sent instead, matching how empty text content is already normalized. (#426)
+* `ChatPosit()` now switches API flavors when `chat.model` is set to a model from the other family: setting a non-Claude model on a Claude-backed chat (or vice versa) swaps the underlying provider so requests use the correct wire format and gateway endpoint, rather than failing with a mismatched request. The `cache` setting is preserved across flavor switches. (Mirrors tidyverse/ellmer#1139.)
+* Anthropic-backed providers (`ChatAnthropic()`, `ChatPosit()`, etc.) now drop thinking blocks that lack a signature (e.g., reasoning emitted by a non-Claude model) when replaying conversation history, instead of failing with `Invalid signature in thinking block`.
 
 
 ## [0.23.0] - 2026-09-04
diff --git a/chatlas/_chat.py b/chatlas/_chat.py
index fb4729d9..9e3b21c8 100644
--- a/chatlas/_chat.py
+++ b/chatlas/_chat.py
@@ -536,7 +536,9 @@ def model(self) -> str:
 
         Setting this updates the model for subsequent requests. The model name
         is not validated, so make sure it's a valid model for the chat's
-        provider.
+        provider. Note that some providers (e.g., `ChatPosit()`) dispatch to a
+        different API flavor based on the model name; setting a model from
+        another flavor swaps out the underlying provider accordingly.
 
         Returns
         -------
@@ -547,7 +549,7 @@ def model(self) -> str:
 
     @model.setter
     def model(self, value: str):
-        self.provider.model = value
+        self.provider = self.provider.set_model(value)
 
     @property
     def conversation_id(self) -> str | None:
diff --git a/chatlas/_provider.py b/chatlas/_provider.py
index f30eea97..aab38d8d 100644
--- a/chatlas/_provider.py
+++ b/chatlas/_provider.py
@@ -197,6 +197,24 @@ def model(self):
     def model(self, value: str):
         self._model = value
 
+    def set_model(self, value: str) -> "Provider[Any, Any]":
+        """
+        Set the model used by the provider.
+
+        The default implementation just updates the model name and returns
+        the provider unchanged. Providers that dispatch to a different API
+        flavor based on the model name (e.g., Posit AI) override this to
+        return a different provider instance when the new model belongs to
+        another flavor.
+
+        Returns
+        -------
+        Provider
+            The provider to use for subsequent requests (possibly `self`).
+        """
+        self._model = value
+        return self
+
     def close(self) -> None:
         """
         Release resources held by this provider (e.g., HTTP connection pools,
diff --git a/chatlas/_provider_anthropic.py b/chatlas/_provider_anthropic.py
index dd304aba..259c86cc 100644
--- a/chatlas/_provider_anthropic.py
+++ b/chatlas/_provider_anthropic.py
@@ -894,8 +894,11 @@ def _as_message_params(self, turns: Sequence[Turn]) -> list["MessageParam"]:
             content = [
                 self._as_content_block(c)
                 for c in turn.contents
-                if not isinstance(c, PROVIDER_ANNOTATION_TYPES)
-                or anthropic_replayable(c)
+                if (
+                    not isinstance(c, PROVIDER_ANNOTATION_TYPES)
+                    or anthropic_replayable(c)
+                )
+                and self._is_replayable_thinking(c)
             ]
 
             # Drop empty assistant turns to avoid an API error
@@ -915,6 +918,15 @@ def _as_message_params(self, turns: Sequence[Turn]) -> list["MessageParam"]:
             messages.append({"role": role, "content": content})
         return messages
 
+    @staticmethod
+    def _is_replayable_thinking(content: Content) -> bool:
+        # Thinking blocks without a signature (e.g., reasoning emitted by a
+        # non-Claude model, replayed after a provider switch) are rejected by
+        # the API ("Invalid `signature` in `thinking` block"), so drop them.
+        if not isinstance(content, ContentThinking):
+            return True
+        return bool((content.extra or {}).get("signature"))
+
     @staticmethod
     def _as_content_block(content: Content) -> "ContentBlockParam":
         if isinstance(content, ContentText):
diff --git a/chatlas/_provider_posit.py b/chatlas/_provider_posit.py
index 3e694eb0..90e367a5 100644
--- a/chatlas/_provider_posit.py
+++ b/chatlas/_provider_posit.py
@@ -309,6 +309,19 @@ def __init__(
             ),
         )
 
+    def set_model(
+        self, value: str
+    ) -> "PositAnthropicProvider | PositOpenAIProvider":
+        if value.startswith("claude"):
+            self._model = value
+            return self
+        return PositOpenAIProvider(
+            base_url=self._gateway_base_url,
+            model=value,
+            credentials=self._credentials,
+            cache=self._cache,
+        )
+
     def list_models(self) -> list[ModelInfo]:
         return list_models_posit(self._gateway_base_url, self._credentials)
 
@@ -320,12 +333,16 @@ def __init__(
         base_url: str,
         model: str,
         credentials: Callable[[], str],
+        cache: Literal["5m", "1h", "none"] = "5m",
         name: str = "Posit",
     ):
         super().__init__(model=model, api_key="not-used", name=name)
 
         self._gateway_base_url = base_url.rstrip("/")
         self._credentials = credentials
+        # Inert on this flavor (caching is Claude-only); stored so the
+        # setting survives a round trip through `set_model()`.
+        self._cache = cache
 
         auth = PositHttpx2Auth(credentials)
         flavor_base_url = f"{self._gateway_base_url}/openai/v1"
@@ -349,6 +366,19 @@ def __init__(
             ),
         )
 
+    def set_model(
+        self, value: str
+    ) -> "PositAnthropicProvider | PositOpenAIProvider":
+        if value.startswith("claude"):
+            return PositAnthropicProvider(
+                base_url=self._gateway_base_url,
+                model=value,
+                credentials=self._credentials,
+                cache=self._cache,
+            )
+        self._model = value
+        return self
+
     def list_models(self) -> list[ModelInfo]:
         return list_models_posit(self._gateway_base_url, self._credentials)
 
@@ -445,6 +475,7 @@ def ChatPosit(
             base_url=base_url,
             model=model,
             credentials=token_provider,
+            cache=cache,
         )
 
     return Chat(provider=provider, system_prompt=system_prompt)
diff --git a/tests/test_provider_anthropic.py b/tests/test_provider_anthropic.py
index d878b2fa..2710e90b 100644
--- a/tests/test_provider_anthropic.py
+++ b/tests/test_provider_anthropic.py
@@ -1145,3 +1145,41 @@ def test_anthropic_value_cost_prices_fallback_at_serving_models_rate():
 
     # opus-4-8 rates ($5/$25 per 1M), not fable-5's ($10/$50).
     assert cost == pytest.approx((1000 * 5 + 50 * 25) / 1e6)
+
+
+def test_anthropic_message_params_drop_unsigned_thinking_blocks():
+    from chatlas._content import ContentText, ContentThinking
+
+    provider = AnthropicProvider(model="claude-sonnet-4-5", api_key="dummy")
+
+    # Thinking without a signature (e.g., reasoning from a non-Claude model,
+    # replayed after a provider switch) can't be sent to the API, so it's
+    # dropped; the text content of the turn is kept.
+    turns = [
+        UserTurn("hi"),
+        AssistantTurn(
+            [ContentThinking(thinking="hmm"), ContentText(text="hello")]
+        ),
+        UserTurn("again"),
+    ]
+    messages = provider._as_message_params(turns)
+    assert messages[1]["content"] == [{"text": "hello", "type": "text"}]
+
+    # Signed thinking (Claude's own) is preserved.
+    turns[1] = AssistantTurn(
+        [
+            ContentThinking(thinking="hmm", extra={"signature": "sig"}),
+            ContentText(text="hello"),
+        ]
+    )
+    messages = provider._as_message_params(turns)
+    assert messages[1]["content"][0] == {
+        "type": "thinking",
+        "thinking": "hmm",
+        "signature": "sig",
+    }
+
+    # A turn consisting solely of unsigned thinking is dropped entirely.
+    turns[1] = AssistantTurn([ContentThinking(thinking="hmm")])
+    messages = provider._as_message_params(turns)
+    assert len(messages) == 2
diff --git a/tests/test_provider_posit.py b/tests/test_provider_posit.py
index e780c661..6d71fac5 100644
--- a/tests/test_provider_posit.py
+++ b/tests/test_provider_posit.py
@@ -537,3 +537,78 @@ def test_chat_posit_importable_from_package_root():
     import chatlas
 
     assert chatlas.ChatPosit is ChatPosit
+
+
+def test_chat_posit_set_model_switches_from_anthropic_to_openai():
+    chat = ChatPosit(model="claude-sonnet-4-6", credentials=lambda: "test-token")
+    chat.model = "qwen3-8b"
+
+    assert isinstance(chat.provider, PositOpenAIProvider)
+    assert chat.provider.model == "qwen3-8b"
+    assert (
+        str(chat.provider._client.base_url).rstrip("/")
+        == "https://gateway.posit.ai/openai/v1"
+    )
+
+
+def test_chat_posit_set_model_switches_from_openai_to_anthropic():
+    chat = ChatPosit(model="qwen3-8b", credentials=lambda: "test-token")
+    chat.model = "claude-sonnet-4-6"
+
+    assert isinstance(chat.provider, PositAnthropicProvider)
+    assert chat.provider.model == "claude-sonnet-4-6"
+    assert (
+        str(chat.provider._client.base_url).rstrip("/")
+        == "https://gateway.posit.ai/anthropic"
+    )
+
+
+def test_chat_posit_set_model_within_family_is_a_noop():
+    chat = ChatPosit(model="claude-sonnet-4-6", credentials=lambda: "test-token")
+    provider = chat.provider
+
+    chat.model = "claude-opus-4-1"
+
+    assert chat.provider is provider
+    assert chat.provider.model == "claude-opus-4-1"
+
+    chat = ChatPosit(model="qwen3-8b", credentials=lambda: "test-token")
+    provider = chat.provider
+
+    chat.model = "gpt-4o"
+
+    assert chat.provider is provider
+    assert chat.provider.model == "gpt-4o"
+
+
+def test_chat_posit_set_model_carries_over_credentials():
+    chat = ChatPosit(model="claude-sonnet-4-6", credentials=lambda: "test-token")
+    chat.model = "qwen3-8b"
+    assert chat.provider._credentials() == "test-token"
+
+    chat.model = "claude-sonnet-4-6"
+    assert chat.provider._credentials() == "test-token"
+
+
+def test_chat_posit_set_model_preserves_cache_across_family_switches():
+    chat = ChatPosit(
+        model="claude-sonnet-4-6", cache="1h", credentials=lambda: "test-token"
+    )
+
+    chat.model = "qwen3-8b"
+    assert isinstance(chat.provider, PositOpenAIProvider)
+
+    chat.model = "claude-sonnet-4-6"
+    assert isinstance(chat.provider, PositAnthropicProvider)
+    assert chat.provider._cache == "1h"
+
+
+def test_chat_posit_set_model_openai_start_honors_cache_on_switch():
+    chat = ChatPosit(
+        model="qwen3-8b", cache="none", credentials=lambda: "test-token"
+    )
+
+    chat.model = "claude-sonnet-4-6"
+
+    assert isinstance(chat.provider, PositAnthropicProvider)
+    assert chat.provider._cache == "none"

From f9dfa79517e226d25ece87d42298dd3d2d72c6c7 Mon Sep 17 00:00:00 2001
From: Carson <cpsievert1@gmail.com>
Date: Tue, 8 Sep 2026 11:11:31 -0500
Subject: [PATCH 2/7] chore: fix pyright errors and match ruff format in
 set_model hook

---
 chatlas/_provider.py       |  2 +-
 chatlas/_provider_posit.py | 10 +++-------
 2 files changed, 4 insertions(+), 8 deletions(-)

diff --git a/chatlas/_provider.py b/chatlas/_provider.py
index aab38d8d..eac98e85 100644
--- a/chatlas/_provider.py
+++ b/chatlas/_provider.py
@@ -197,7 +197,7 @@ def model(self):
     def model(self, value: str):
         self._model = value
 
-    def set_model(self, value: str) -> "Provider[Any, Any]":
+    def set_model(self, value: str) -> "Provider[Any, Any, Any, Any]":
         """
         Set the model used by the provider.
 
diff --git a/chatlas/_provider_posit.py b/chatlas/_provider_posit.py
index 90e367a5..20cc5733 100644
--- a/chatlas/_provider_posit.py
+++ b/chatlas/_provider_posit.py
@@ -309,9 +309,7 @@ def __init__(
             ),
         )
 
-    def set_model(
-        self, value: str
-    ) -> "PositAnthropicProvider | PositOpenAIProvider":
+    def set_model(self, value: str) -> "PositAnthropicProvider | PositOpenAIProvider":
         if value.startswith("claude"):
             self._model = value
             return self
@@ -342,7 +340,7 @@ def __init__(
         self._credentials = credentials
         # Inert on this flavor (caching is Claude-only); stored so the
         # setting survives a round trip through `set_model()`.
-        self._cache = cache
+        self._cache: Literal["5m", "1h", "none"] = cache
 
         auth = PositHttpx2Auth(credentials)
         flavor_base_url = f"{self._gateway_base_url}/openai/v1"
@@ -366,9 +364,7 @@ def __init__(
             ),
         )
 
-    def set_model(
-        self, value: str
-    ) -> "PositAnthropicProvider | PositOpenAIProvider":
+    def set_model(self, value: str) -> "PositAnthropicProvider | PositOpenAIProvider":
         if value.startswith("claude"):
             return PositAnthropicProvider(
                 base_url=self._gateway_base_url,

From d53d1eab1998a2e35b5512e543bf2ffc913b1f27 Mon Sep 17 00:00:00 2001
From: "github-actions[bot]" <github-actions[bot]@users.noreply.github.com>
Date: Tue, 8 Sep 2026 16:15:00 +0000
Subject: [PATCH 3/7] chore: update provider types and pricing data

---
 chatlas/data/prices.json           | 2263 ++++++++++++++++++++++++----
 chatlas/types/anthropic/_submit.py |    1 +
 2 files changed, 2004 insertions(+), 260 deletions(-)

diff --git a/chatlas/data/prices.json b/chatlas/data/prices.json
index e5cb4e40..e8c153f9 100644
--- a/chatlas/data/prices.json
+++ b/chatlas/data/prices.json
@@ -1,7 +1,7 @@
 {
   "schema_version": 1,
   "min_ellmer_version": "0.5.0",
-  "updated_at": "2026-09-02T12:13:24Z",
+  "updated_at": "2026-09-07T11:15:25Z",
   "data": [
     {
       "provider": "AWS/Bedrock",
@@ -2054,7 +2054,7 @@
       "provider": "AWS/Bedrock",
       "model": "qwen.qwen3-coder-480b-a35b-v1:0",
       "variant": "",
-      "input": 0.22,
+      "input": 0.45,
       "output": 1.8
     },
     {
@@ -2360,6 +2360,14 @@
       "output": 6,
       "cached_input": 0.12
     },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-east-1/anthropic.claude-opus-4-8",
+      "variant": "",
+      "input": 6,
+      "output": 30,
+      "cached_input": 0.6
+    },
     {
       "provider": "AWS/Bedrock",
       "model": "us-gov-east-1/anthropic.claude-sonnet-4-5-20250929-v1:0",
@@ -2368,6 +2376,14 @@
       "output": 18,
       "cached_input": 0.36
     },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-east-1/anthropic.claude-sonnet-5",
+      "variant": "",
+      "input": 2.4,
+      "output": 12,
+      "cached_input": 0.24
+    },
     {
       "provider": "AWS/Bedrock",
       "model": "us-gov-east-1/claude-sonnet-4-5-20250929-v1:0",
@@ -2390,6 +2406,49 @@
       "input": 0.3,
       "output": 2.65
     },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-east-1/nvidia.nemotron-nano-12b-v2",
+      "variant": "",
+      "input": 0.24,
+      "output": 0.72
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-east-1/nvidia.nemotron-nano-3-30b",
+      "variant": "",
+      "input": 0.072,
+      "output": 0.288
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-east-1/nvidia.nemotron-super-3-120b",
+      "variant": "",
+      "input": 0.18,
+      "output": 0.78
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-east-1/openai.gpt-5.4",
+      "variant": "",
+      "input": 3.3,
+      "output": 19.8,
+      "cached_input": 0.33
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-east-1/openai.gpt-oss-120b-1:0",
+      "variant": "",
+      "input": 0.18,
+      "output": 0.72
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-east-1/openai.gpt-oss-20b-1:0",
+      "variant": "",
+      "input": 0.084,
+      "output": 0.36
+    },
     {
       "provider": "AWS/Bedrock",
       "model": "us-gov-west-1/amazon.nova-pro-v1:0",
@@ -2464,6 +2523,14 @@
       "output": 6,
       "cached_input": 0.12
     },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-west-1/anthropic.claude-opus-4-8",
+      "variant": "",
+      "input": 6,
+      "output": 30,
+      "cached_input": 0.6
+    },
     {
       "provider": "AWS/Bedrock",
       "model": "us-gov-west-1/anthropic.claude-sonnet-4-5-20250929-v1:0",
@@ -2472,6 +2539,14 @@
       "output": 18,
       "cached_input": 0.36
     },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-west-1/anthropic.claude-sonnet-5",
+      "variant": "",
+      "input": 2.4,
+      "output": 12,
+      "cached_input": 0.24
+    },
     {
       "provider": "AWS/Bedrock",
       "model": "us-gov-west-1/claude-sonnet-4-5-20250929-v1:0",
@@ -2492,7 +2567,98 @@
       "model": "us-gov-west-1/meta.llama3-8b-instruct-v1:0",
       "variant": "",
       "input": 0.3,
-      "output": 2.65
+      "output": 0.6
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-west-1/nvidia.nemotron-nano-12b-v2",
+      "variant": "",
+      "input": 0.24,
+      "output": 0.72
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-west-1/nvidia.nemotron-nano-3-30b",
+      "variant": "",
+      "input": 0.072,
+      "output": 0.288
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-west-1/nvidia.nemotron-super-3-120b",
+      "variant": "",
+      "input": 0.18,
+      "output": 0.78
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-west-1/openai.gpt-5.4",
+      "variant": "",
+      "input": 3.3,
+      "output": 19.8,
+      "cached_input": 0.33
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-west-1/openai.gpt-5.6-luna",
+      "variant": "",
+      "input": 0.264,
+      "output": 1.584,
+      "cached_input": 0.0264
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-west-1/openai.gpt-5.6-luna",
+      "variant": "above_272k_tokens",
+      "input": 0.528,
+      "output": 2.376,
+      "cached_input": 0.0528
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-west-1/openai.gpt-5.6-terra",
+      "variant": "",
+      "input": 2.64,
+      "output": 15.84,
+      "cached_input": 0.264
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-west-1/openai.gpt-5.6-terra",
+      "variant": "above_272k_tokens",
+      "input": 5.28,
+      "output": 23.76,
+      "cached_input": 0.528
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-west-1/openai.gpt-oss-120b-1:0",
+      "variant": "",
+      "input": 0.18,
+      "output": 0.72
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-west-1/openai.gpt-oss-20b-1:0",
+      "variant": "",
+      "input": 0.084,
+      "output": 0.36
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov-west-1/xai.grok-4.3",
+      "variant": "",
+      "input": 1.5,
+      "output": 3,
+      "cached_input": 0.24
+    },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov.anthropic.claude-opus-4-8",
+      "variant": "",
+      "input": 6,
+      "output": 30,
+      "cached_input": 0.6
     },
     {
       "provider": "AWS/Bedrock",
@@ -2510,6 +2676,14 @@
       "output": 27,
       "cached_input": 0.72
     },
+    {
+      "provider": "AWS/Bedrock",
+      "model": "us-gov.anthropic.claude-sonnet-5",
+      "variant": "",
+      "input": 2.4,
+      "output": 12,
+      "cached_input": 0.24
+    },
     {
       "provider": "AWS/Bedrock",
       "model": "us-west-1/meta.llama3-70b-instruct-v1:0",
@@ -3184,6 +3358,14 @@
       "output": 50,
       "cached_input": 1
     },
+    {
+      "provider": "Anthropic",
+      "model": "claude-mythos-5-1",
+      "variant": "",
+      "input": 10,
+      "output": 50,
+      "cached_input": 0.25
+    },
     {
       "provider": "Anthropic",
       "model": "claude-mythos-preview",
@@ -4768,6 +4950,22 @@
       "output": 24,
       "cached_input": 0.4
     },
+    {
+      "provider": "Azure/OpenAI",
+      "model": "gpt-6-astra",
+      "variant": "",
+      "input": 10,
+      "output": 50,
+      "cached_input": 1
+    },
+    {
+      "provider": "Azure/OpenAI",
+      "model": "gpt-6-astra",
+      "variant": "above_272k_tokens",
+      "input": 20,
+      "output": 75,
+      "cached_input": 2
+    },
     {
       "provider": "Azure/OpenAI",
       "model": "gpt-audio-1.5-2026-02-23",
@@ -4829,7 +5027,6 @@
       "model": "gpt-image-2",
       "variant": "",
       "input": 5,
-      "output": 10,
       "cached_input": 1.25
     },
     {
@@ -4837,7 +5034,6 @@
       "model": "gpt-image-2-2026-04-21",
       "variant": "",
       "input": 5,
-      "output": 10,
       "cached_input": 1.25
     },
     {
@@ -4848,6 +5044,30 @@
       "output": 16,
       "cached_input": 4
     },
+    {
+      "provider": "Azure/OpenAI",
+      "model": "gpt-realtime-2",
+      "variant": "",
+      "input": 4,
+      "output": 24,
+      "cached_input": 0.4
+    },
+    {
+      "provider": "Azure/OpenAI",
+      "model": "gpt-realtime-2.1",
+      "variant": "",
+      "input": 4,
+      "output": 24,
+      "cached_input": 0.4
+    },
+    {
+      "provider": "Azure/OpenAI",
+      "model": "gpt-realtime-2.1-mini",
+      "variant": "",
+      "input": 0.6,
+      "output": 2.4,
+      "cached_input": 0.06
+    },
     {
       "provider": "Azure/OpenAI",
       "model": "gpt-realtime-2025-08-28",
@@ -5039,6 +5259,36 @@
       "input": 0.1,
       "output": 0
     },
+    {
+      "provider": "Azure/OpenAI",
+      "model": "us-gov/gpt-5.1",
+      "variant": "",
+      "input": 1.71875,
+      "output": 13.75,
+      "cached_input": 0.171875
+    },
+    {
+      "provider": "Azure/OpenAI",
+      "model": "us-gov/o3-mini",
+      "variant": "",
+      "input": 1.513,
+      "output": 6.05,
+      "cached_input": 0.757
+    },
+    {
+      "provider": "Azure/OpenAI",
+      "model": "us-gov/text-embedding-3-large",
+      "variant": "",
+      "input": 0.163,
+      "output": 0
+    },
+    {
+      "provider": "Azure/OpenAI",
+      "model": "us-gov/text-embedding-3-small",
+      "variant": "",
+      "input": 0.025,
+      "output": 0
+    },
     {
       "provider": "Azure/OpenAI",
       "model": "us/gpt-4.1-2025-04-14",
@@ -5395,6 +5645,22 @@
       "output": 26.4,
       "cached_input": 0.44
     },
+    {
+      "provider": "Azure/OpenAI",
+      "model": "us/gpt-6-astra",
+      "variant": "",
+      "input": 11,
+      "output": 55,
+      "cached_input": 1.1
+    },
+    {
+      "provider": "Azure/OpenAI",
+      "model": "us/gpt-6-astra",
+      "variant": "above_272k_tokens",
+      "input": 22,
+      "output": 82.5,
+      "cached_input": 2.2
+    },
     {
       "provider": "Azure/OpenAI",
       "model": "us/o1-2024-12-17",
@@ -6045,6 +6311,37 @@
       "output": 6.75,
       "cached_input": 0.135
     },
+    {
+      "provider": "Google/Gemini",
+      "model": "gemini-3.8-flash",
+      "variant": "",
+      "input": 0.75,
+      "output": 3.75,
+      "cached_input": 0.075
+    },
+    {
+      "provider": "Google/Gemini",
+      "model": "gemini-3.8-flash",
+      "variant": "batches",
+      "input": 0.375,
+      "output": 1.875
+    },
+    {
+      "provider": "Google/Gemini",
+      "model": "gemini-3.8-flash",
+      "variant": "flex",
+      "input": 0.375,
+      "output": 1.875,
+      "cached_input": 0.0375
+    },
+    {
+      "provider": "Google/Gemini",
+      "model": "gemini-3.8-flash",
+      "variant": "priority",
+      "input": 1.35,
+      "output": 6.75,
+      "cached_input": 0.135
+    },
     {
       "provider": "Google/Gemini",
       "model": "gemini-embedding-001",
@@ -6657,30 +6954,61 @@
     },
     {
       "provider": "Google/Vertex",
-      "model": "gemini-live-2.5-flash-native-audio",
-      "variant": "",
-      "input": 0.5,
-      "output": 2
-    },
-    {
-      "provider": "Google/Vertex",
-      "model": "gemini-live-2.5-flash-preview-native-audio-09-2025",
+      "model": "gemini-3.8-flash",
       "variant": "",
-      "input": 0.5,
-      "output": 2,
+      "input": 0.75,
+      "output": 3.75,
       "cached_input": 0.075
     },
     {
       "provider": "Google/Vertex",
-      "model": "gemini-omni-flash-preview",
-      "variant": "",
-      "input": 1.5,
-      "output": 9
+      "model": "gemini-3.8-flash",
+      "variant": "batches",
+      "input": 0.375,
+      "output": 1.875
     },
     {
       "provider": "Google/Vertex",
-      "model": "gemini-robotics-er-1.5-preview",
-      "variant": "",
+      "model": "gemini-3.8-flash",
+      "variant": "flex",
+      "input": 0.375,
+      "output": 1.875,
+      "cached_input": 0.0375
+    },
+    {
+      "provider": "Google/Vertex",
+      "model": "gemini-3.8-flash",
+      "variant": "priority",
+      "input": 1.35,
+      "output": 6.75,
+      "cached_input": 0.135
+    },
+    {
+      "provider": "Google/Vertex",
+      "model": "gemini-live-2.5-flash-native-audio",
+      "variant": "",
+      "input": 0.5,
+      "output": 2
+    },
+    {
+      "provider": "Google/Vertex",
+      "model": "gemini-live-2.5-flash-preview-native-audio-09-2025",
+      "variant": "",
+      "input": 0.5,
+      "output": 2,
+      "cached_input": 0.075
+    },
+    {
+      "provider": "Google/Vertex",
+      "model": "gemini-omni-flash-preview",
+      "variant": "",
+      "input": 1.5,
+      "output": 9
+    },
+    {
+      "provider": "Google/Vertex",
+      "model": "gemini-robotics-er-1.5-preview",
+      "variant": "",
       "input": 0.3,
       "output": 2.5,
       "cached_input": 0
@@ -7076,8 +7404,9 @@
       "provider": "Mistral",
       "model": "magistral-medium-latest",
       "variant": "",
-      "input": 2,
-      "output": 5
+      "input": 1.5,
+      "output": 7.5,
+      "cached_input": 0.15
     },
     {
       "provider": "Mistral",
@@ -7097,22 +7426,25 @@
       "provider": "Mistral",
       "model": "magistral-small-latest",
       "variant": "",
-      "input": 0.5,
-      "output": 1.5
+      "input": 0.15,
+      "output": 0.6,
+      "cached_input": 0.015
     },
     {
       "provider": "Mistral",
       "model": "ministral-14b-2512",
       "variant": "",
       "input": 0.2,
-      "output": 0.2
+      "output": 0.2,
+      "cached_input": 0.02
     },
     {
       "provider": "Mistral",
       "model": "ministral-14b-latest",
       "variant": "",
       "input": 0.2,
-      "output": 0.2
+      "output": 0.2,
+      "cached_input": 0.02
     },
     {
       "provider": "Mistral",
@@ -7143,14 +7475,16 @@
       "model": "ministral-3b-2512",
       "variant": "",
       "input": 0.1,
-      "output": 0.1
+      "output": 0.1,
+      "cached_input": 0.01
     },
     {
       "provider": "Mistral",
       "model": "ministral-3b-latest",
       "variant": "",
       "input": 0.1,
-      "output": 0.1
+      "output": 0.1,
+      "cached_input": 0.01
     },
     {
       "provider": "Mistral",
@@ -7252,8 +7586,9 @@
       "provider": "Mistral",
       "model": "mistral-medium",
       "variant": "",
-      "input": 2.7,
-      "output": 8.1
+      "input": 1.5,
+      "output": 7.5,
+      "cached_input": 0.15
     },
     {
       "provider": "Mistral",
@@ -8640,6 +8975,14 @@
       "output": 22.5,
       "cached_input": 0.5
     },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-5.4",
+      "variant": "above_272k_tokens_flex",
+      "input": 2.5,
+      "output": 11.25,
+      "cached_input": 0.25
+    },
     {
       "provider": "OpenAI",
       "model": "gpt-5.4",
@@ -8679,6 +9022,14 @@
       "output": 22.5,
       "cached_input": 0.5
     },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-5.4-2026-03-05",
+      "variant": "above_272k_tokens_flex",
+      "input": 2.5,
+      "output": 11.25,
+      "cached_input": 0.25
+    },
     {
       "provider": "OpenAI",
       "model": "gpt-5.4-2026-03-05",
@@ -8826,6 +9177,13 @@
       "output": 270,
       "cached_input": 6
     },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-5.4-pro",
+      "variant": "above_272k_tokens_flex",
+      "input": 30,
+      "output": 135
+    },
     {
       "provider": "OpenAI",
       "model": "gpt-5.4-pro",
@@ -8856,6 +9214,13 @@
       "output": 270,
       "cached_input": 6
     },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-5.4-pro-2026-03-05",
+      "variant": "above_272k_tokens_flex",
+      "input": 30,
+      "output": 135
+    },
     {
       "provider": "OpenAI",
       "model": "gpt-5.4-pro-2026-03-05",
@@ -8886,6 +9251,14 @@
       "output": 45,
       "cached_input": 1
     },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-5.5",
+      "variant": "above_272k_tokens_flex",
+      "input": 5,
+      "output": 22.5,
+      "cached_input": 0.5
+    },
     {
       "provider": "OpenAI",
       "model": "gpt-5.5",
@@ -8905,9 +9278,9 @@
       "provider": "OpenAI",
       "model": "gpt-5.5",
       "variant": "priority",
-      "input": 10,
-      "output": 60,
-      "cached_input": 1
+      "input": 12.5,
+      "output": 75,
+      "cached_input": 1.25
     },
     {
       "provider": "OpenAI",
@@ -8925,6 +9298,14 @@
       "output": 45,
       "cached_input": 1
     },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-5.5-2026-04-23",
+      "variant": "above_272k_tokens_flex",
+      "input": 5,
+      "output": 22.5,
+      "cached_input": 0.5
+    },
     {
       "provider": "OpenAI",
       "model": "gpt-5.5-2026-04-23",
@@ -8944,9 +9325,9 @@
       "provider": "OpenAI",
       "model": "gpt-5.5-2026-04-23",
       "variant": "priority",
-      "input": 10,
-      "output": 60,
-      "cached_input": 1
+      "input": 12.5,
+      "output": 75,
+      "cached_input": 1.25
     },
     {
       "provider": "OpenAI",
@@ -9032,6 +9413,14 @@
       "output": 15,
       "cached_input": 0.4
     },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-5.6",
+      "variant": "above_272k_tokens_priority",
+      "input": 16,
+      "output": 60,
+      "cached_input": 1.6
+    },
     {
       "provider": "OpenAI",
       "model": "gpt-5.6",
@@ -9095,6 +9484,14 @@
       "output": 0.9,
       "cached_input": 0.02
     },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-5.6-luna",
+      "variant": "above_272k_tokens_priority",
+      "input": 0.8,
+      "output": 3.6,
+      "cached_input": 0.08
+    },
     {
       "provider": "OpenAI",
       "model": "gpt-5.6-luna",
@@ -9142,6 +9539,14 @@
       "output": 15,
       "cached_input": 0.4
     },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-5.6-sol",
+      "variant": "above_272k_tokens_priority",
+      "input": 16,
+      "output": 60,
+      "cached_input": 1.6
+    },
     {
       "provider": "OpenAI",
       "model": "gpt-5.6-sol",
@@ -9189,6 +9594,14 @@
       "output": 9,
       "cached_input": 0.2
     },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-5.6-terra",
+      "variant": "above_272k_tokens_priority",
+      "input": 8,
+      "output": 36,
+      "cached_input": 0.8
+    },
     {
       "provider": "OpenAI",
       "model": "gpt-5.6-terra",
@@ -9212,6 +9625,61 @@
       "output": 24,
       "cached_input": 0.4
     },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-6-astra",
+      "variant": "",
+      "input": 10,
+      "output": 50,
+      "cached_input": 1
+    },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-6-astra",
+      "variant": "above_272k_tokens",
+      "input": 20,
+      "output": 75,
+      "cached_input": 2
+    },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-6-astra",
+      "variant": "above_272k_tokens_flex",
+      "input": 10,
+      "output": 37.5,
+      "cached_input": 1
+    },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-6-astra",
+      "variant": "above_272k_tokens_priority",
+      "input": 40,
+      "output": 150,
+      "cached_input": 4
+    },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-6-astra",
+      "variant": "batches",
+      "input": 5,
+      "output": 25
+    },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-6-astra",
+      "variant": "flex",
+      "input": 5,
+      "output": 25,
+      "cached_input": 0.5
+    },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-6-astra",
+      "variant": "priority",
+      "input": 20,
+      "output": 100,
+      "cached_input": 2
+    },
     {
       "provider": "OpenAI",
       "model": "gpt-audio",
@@ -9254,6 +9722,38 @@
       "input": 0.6,
       "output": 2.4
     },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-daybreak-blue-latest",
+      "variant": "",
+      "input": 4,
+      "output": 20,
+      "cached_input": 0.4
+    },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-daybreak-blue-latest",
+      "variant": "above_272k_tokens",
+      "input": 8,
+      "output": 30,
+      "cached_input": 0.8
+    },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-daybreak-red-latest",
+      "variant": "",
+      "input": 12.5,
+      "output": 75,
+      "cached_input": 1.25
+    },
+    {
+      "provider": "OpenAI",
+      "model": "gpt-daybreak-red-latest",
+      "variant": "above_272k_tokens",
+      "input": 25,
+      "output": 112.5,
+      "cached_input": 2.5
+    },
     {
       "provider": "OpenAI",
       "model": "gpt-image-1",
@@ -9289,7 +9789,6 @@
       "model": "gpt-image-2",
       "variant": "",
       "input": 5,
-      "output": 10,
       "cached_input": 1.25
     },
     {
@@ -9297,7 +9796,6 @@
       "model": "gpt-image-2-2026-04-21",
       "variant": "",
       "input": 5,
-      "output": 10,
       "cached_input": 1.25
     },
     {
@@ -9321,7 +9819,7 @@
       "model": "gpt-realtime-2",
       "variant": "",
       "input": 4,
-      "output": 16,
+      "output": 24,
       "cached_input": 0.4
     },
     {
@@ -9685,6 +10183,22 @@
       "input": 3,
       "output": 15
     },
+    {
+      "provider": "OpenRouter",
+      "model": "anthropic/claude-fable-5",
+      "variant": "",
+      "input": 10,
+      "output": 50,
+      "cached_input": 1
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "anthropic/claude-fable-5.1",
+      "variant": "",
+      "input": 10,
+      "output": 50,
+      "cached_input": 0.25
+    },
     {
       "provider": "OpenRouter",
       "model": "anthropic/claude-haiku-4.5",
@@ -9733,6 +10247,14 @@
       "output": 25,
       "cached_input": 0.5
     },
+    {
+      "provider": "OpenRouter",
+      "model": "anthropic/claude-opus-4.8",
+      "variant": "",
+      "input": 5,
+      "output": 25,
+      "cached_input": 0.5
+    },
     {
       "provider": "OpenRouter",
       "model": "anthropic/claude-opus-5",
@@ -9789,6 +10311,14 @@
       "output": 22.5,
       "cached_input": 0.6
     },
+    {
+      "provider": "OpenRouter",
+      "model": "anthropic/claude-sonnet-5",
+      "variant": "",
+      "input": 2,
+      "output": 10,
+      "cached_input": 0.2
+    },
     {
       "provider": "OpenRouter",
       "model": "bytedance/ui-tars-1.5-7b",
@@ -9800,15 +10330,15 @@
       "provider": "OpenRouter",
       "model": "deepseek/deepseek-chat",
       "variant": "",
-      "input": 0.14,
-      "output": 0.28
+      "input": 0.32,
+      "output": 0.89
     },
     {
       "provider": "OpenRouter",
       "model": "deepseek/deepseek-chat-v3-0324",
       "variant": "",
-      "input": 0.14,
-      "output": 0.28
+      "input": 0.25,
+      "output": 1
     },
     {
       "provider": "OpenRouter",
@@ -9827,8 +10357,8 @@
       "provider": "OpenRouter",
       "model": "deepseek/deepseek-r1",
       "variant": "",
-      "input": 0.55,
-      "output": 2.19
+      "input": 0.7,
+      "output": 2.5
     },
     {
       "provider": "OpenRouter",
@@ -9849,11 +10379,26 @@
       "variant": "cache_hit",
       "input": 0.14
     },
+    {
+      "provider": "OpenRouter",
+      "model": "deepseek/deepseek-r1-distill-llama-70b",
+      "variant": "",
+      "input": 0.8,
+      "output": 0.8
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "deepseek/deepseek-v3.1-terminus",
+      "variant": "",
+      "input": 0.27,
+      "output": 1,
+      "cached_input": 0.135
+    },
     {
       "provider": "OpenRouter",
       "model": "deepseek/deepseek-v3.2",
       "variant": "",
-      "input": 0.28,
+      "input": 0.269,
       "output": 0.4
     },
     {
@@ -9866,8 +10411,8 @@
       "provider": "OpenRouter",
       "model": "deepseek/deepseek-v3.2-exp",
       "variant": "",
-      "input": 0.2,
-      "output": 0.4
+      "input": 0.27,
+      "output": 0.41
     },
     {
       "provider": "OpenRouter",
@@ -9875,6 +10420,30 @@
       "variant": "cache_hit",
       "input": 0.02
     },
+    {
+      "provider": "OpenRouter",
+      "model": "deepseek/deepseek-v4-flash",
+      "variant": "",
+      "input": 0.08778,
+      "output": 0.17556,
+      "cached_input": 0.017556
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "deepseek/deepseek-v4-flash-0731",
+      "variant": "",
+      "input": 0.065,
+      "output": 0.18,
+      "cached_input": 0.016
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "deepseek/deepseek-v4-flash-vision-exp",
+      "variant": "",
+      "input": 0.22,
+      "output": 0.66,
+      "cached_input": 0.007
+    },
     {
       "provider": "OpenRouter",
       "model": "deepseek/deepseek-v4-pro",
@@ -9913,14 +10482,64 @@
       "model": "google/gemini-2.5-flash",
       "variant": "",
       "input": 0.3,
-      "output": 2.5
+      "output": 2.5,
+      "cached_input": 0.03
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "google/gemini-2.5-flash-image",
+      "variant": "",
+      "input": 0.3,
+      "output": 2.5,
+      "cached_input": 0.03
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "google/gemini-2.5-flash-lite",
+      "variant": "",
+      "input": 0.1,
+      "output": 0.4,
+      "cached_input": 0.01
     },
     {
       "provider": "OpenRouter",
       "model": "google/gemini-2.5-pro",
       "variant": "",
       "input": 1.25,
-      "output": 10
+      "output": 10,
+      "cached_input": 0.125
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "google/gemini-2.5-pro-preview",
+      "variant": "",
+      "input": 1.25,
+      "output": 10,
+      "cached_input": 0.125
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "google/gemini-2.5-pro-preview",
+      "variant": "above_200k_tokens",
+      "input": 2.5,
+      "output": 15,
+      "cached_input": 0.25
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "google/gemini-2.5-pro-preview-05-06",
+      "variant": "",
+      "input": 1.25,
+      "output": 10,
+      "cached_input": 0.125
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "google/gemini-2.5-pro-preview-05-06",
+      "variant": "above_200k_tokens",
+      "input": 2.5,
+      "output": 15,
+      "cached_input": 0.25
     },
     {
       "provider": "OpenRouter",
@@ -9930,6 +10549,22 @@
       "output": 3,
       "cached_input": 0.05
     },
+    {
+      "provider": "OpenRouter",
+      "model": "google/gemini-3-pro-image",
+      "variant": "",
+      "input": 2,
+      "output": 12,
+      "cached_input": 0.2
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "google/gemini-3-pro-image-preview",
+      "variant": "",
+      "input": 2,
+      "output": 12,
+      "cached_input": 0.2
+    },
     {
       "provider": "OpenRouter",
       "model": "google/gemini-3-pro-preview",
@@ -9953,6 +10588,20 @@
       "input": 1,
       "output": 6
     },
+    {
+      "provider": "OpenRouter",
+      "model": "google/gemini-3.1-flash-image",
+      "variant": "",
+      "input": 0.5,
+      "output": 3
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "google/gemini-3.1-flash-image-preview",
+      "variant": "",
+      "input": 0.5,
+      "output": 3
+    },
     {
       "provider": "OpenRouter",
       "model": "google/gemini-3.1-flash-lite",
@@ -9961,6 +10610,13 @@
       "output": 1.5,
       "cached_input": 0.025
     },
+    {
+      "provider": "OpenRouter",
+      "model": "google/gemini-3.1-flash-lite-image",
+      "variant": "",
+      "input": 0.25,
+      "output": 1.5
+    },
     {
       "provider": "OpenRouter",
       "model": "google/gemini-3.1-flash-lite-preview",
@@ -9987,195 +10643,660 @@
     },
     {
       "provider": "OpenRouter",
-      "model": "gryphe/mythomax-l2-13b",
+      "model": "google/gemini-3.1-pro-preview-customtools",
       "variant": "",
-      "input": 1.875,
-      "output": 1.875
+      "input": 2,
+      "output": 12,
+      "cached_input": 0.2
     },
     {
       "provider": "OpenRouter",
-      "model": "mancer/weaver",
-      "variant": "",
-      "input": 5.625,
-      "output": 5.625
+      "model": "google/gemini-3.1-pro-preview-customtools",
+      "variant": "above_200k_tokens",
+      "input": 4,
+      "output": 18,
+      "cached_input": 0.4
     },
     {
       "provider": "OpenRouter",
-      "model": "meta-llama/llama-3-70b-instruct",
+      "model": "google/gemini-3.5-flash",
       "variant": "",
-      "input": 0.59,
-      "output": 0.79
+      "input": 1.5,
+      "output": 9,
+      "cached_input": 0.15
     },
     {
       "provider": "OpenRouter",
-      "model": "minimax/minimax-m2",
+      "model": "google/gemini-3.5-flash-lite",
       "variant": "",
-      "input": 0.255,
-      "output": 1.02
+      "input": 0.3,
+      "output": 2.5,
+      "cached_input": 0.03
     },
     {
       "provider": "OpenRouter",
-      "model": "minimax/minimax-m2.1",
+      "model": "google/gemini-3.6-flash",
       "variant": "",
-      "input": 0.27,
-      "output": 1.2,
-      "cached_input": 0
+      "input": 0.75,
+      "output": 3.75,
+      "cached_input": 0.075
     },
     {
       "provider": "OpenRouter",
-      "model": "minimax/minimax-m2.5",
+      "model": "google/gemini-3.7-flash",
       "variant": "",
-      "input": 0.3,
-      "output": 1.1,
-      "cached_input": 0.15
+      "input": 0.75,
+      "output": 3.75,
+      "cached_input": 0.075
     },
     {
       "provider": "OpenRouter",
-      "model": "mistralai/devstral-2512",
+      "model": "google/gemini-3.8-flash",
       "variant": "",
-      "input": 0.15,
-      "output": 0.6
+      "input": 0.75,
+      "output": 3.75,
+      "cached_input": 0.075
     },
     {
       "provider": "OpenRouter",
-      "model": "mistralai/ministral-14b-2512",
+      "model": "google/gemma-2-27b-it",
       "variant": "",
-      "input": 0.2,
-      "output": 0.2
+      "input": 0.65,
+      "output": 0.65
     },
     {
       "provider": "OpenRouter",
-      "model": "mistralai/ministral-3b-2512",
+      "model": "google/gemma-3-12b-it",
       "variant": "",
-      "input": 0.1,
-      "output": 0.1
+      "input": 0.05,
+      "output": 0.15
     },
     {
       "provider": "OpenRouter",
-      "model": "mistralai/ministral-8b-2512",
+      "model": "google/gemma-3-27b-it",
       "variant": "",
-      "input": 0.15,
-      "output": 0.15
+      "input": 0.08,
+      "output": 0.45,
+      "cached_input": 0.04
     },
     {
       "provider": "OpenRouter",
-      "model": "mistralai/mistral-7b-instruct",
+      "model": "google/gemma-3-4b-it",
       "variant": "",
-      "input": 0.13,
-      "output": 0.13
+      "input": 0.05,
+      "output": 0.1
     },
     {
       "provider": "OpenRouter",
-      "model": "mistralai/mistral-large",
+      "model": "google/gemma-4-26b-a4b-it",
       "variant": "",
-      "input": 8,
-      "output": 24
+      "input": 0.07,
+      "output": 0.34
     },
     {
       "provider": "OpenRouter",
-      "model": "mistralai/mistral-large-2512",
+      "model": "google/gemma-4-31b-it",
       "variant": "",
-      "input": 0.5,
-      "output": 1.5
+      "input": 0.09,
+      "output": 0.34,
+      "cached_input": 0.05
     },
     {
       "provider": "OpenRouter",
-      "model": "mistralai/mistral-small-3.1-24b-instruct",
+      "model": "gryphe/mythomax-l2-13b",
       "variant": "",
-      "input": 0.1,
-      "output": 0.3
+      "input": 0.06,
+      "output": 0.06
     },
     {
       "provider": "OpenRouter",
-      "model": "mistralai/mistral-small-3.2-24b-instruct",
+      "model": "mancer/weaver",
       "variant": "",
-      "input": 0.1,
-      "output": 0.3
+      "input": 0.4,
+      "output": 0.75
     },
     {
       "provider": "OpenRouter",
-      "model": "mistralai/mixtral-8x22b-instruct",
+      "model": "meta-llama/llama-3-70b-instruct",
       "variant": "",
-      "input": 0.65,
-      "output": 0.65
+      "input": 0.59,
+      "output": 0.79
     },
     {
       "provider": "OpenRouter",
-      "model": "moonshotai/kimi-k2.5",
+      "model": "meta-llama/llama-3.1-70b-instruct",
       "variant": "",
-      "input": 0.6,
-      "output": 3,
-      "cached_input": 0.1
+      "input": 0.4,
+      "output": 0.4
     },
     {
       "provider": "OpenRouter",
-      "model": "nvidia/nemotron-3.5-lightning",
+      "model": "meta-llama/llama-3.1-8b-instruct",
       "variant": "",
       "input": 0.05,
-      "output": 0.2
+      "output": 0.08,
+      "cached_input": 0.025
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-3.5-turbo",
+      "model": "meta-llama/llama-3.2-1b-instruct",
       "variant": "",
-      "input": 1.5,
-      "output": 2
+      "input": 0.027,
+      "output": 0.201
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-3.5-turbo-16k",
+      "model": "meta-llama/llama-3.2-3b-instruct",
       "variant": "",
-      "input": 3,
-      "output": 4
+      "input": 0.05,
+      "output": 0.33
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-4",
+      "model": "meta-llama/llama-3.3-70b-instruct",
       "variant": "",
-      "input": 30,
-      "output": 60
+      "input": 0.1,
+      "output": 0.32
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-4.1",
+      "model": "meta-llama/llama-4-maverick",
       "variant": "",
-      "input": 2,
-      "output": 8,
-      "cached_input": 0.5
+      "input": 0.2,
+      "output": 0.696
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-4.1-mini",
+      "model": "meta-llama/llama-4-scout",
       "variant": "",
-      "input": 0.4,
-      "output": 1.6,
-      "cached_input": 0.1
+      "input": 0.1,
+      "output": 0.3
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-4.1-nano",
+      "model": "meta-llama/llama-guard-4-12b",
       "variant": "",
-      "input": 0.1,
-      "output": 0.4,
-      "cached_input": 0.025
+      "input": 0.18,
+      "output": 0.18
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-4o",
+      "model": "minimax/minimax-01",
       "variant": "",
-      "input": 2.5,
-      "output": 10
+      "input": 0.2,
+      "output": 1.1
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-4o-2024-05-13",
+      "model": "minimax/minimax-m1",
       "variant": "",
-      "input": 5,
-      "output": 15
+      "input": 0.55,
+      "output": 2.2
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-5",
+      "model": "minimax/minimax-m2",
+      "variant": "",
+      "input": 0.255,
+      "output": 1.02
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "minimax/minimax-m2-her",
+      "variant": "",
+      "input": 0.3,
+      "output": 1.2,
+      "cached_input": 0.03
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "minimax/minimax-m2.1",
+      "variant": "",
+      "input": 0.3,
+      "output": 1.2,
+      "cached_input": 0.03
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "minimax/minimax-m2.5",
+      "variant": "",
+      "input": 0.27,
+      "output": 1.08,
+      "cached_input": 0.027
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "minimax/minimax-m2.7",
+      "variant": "",
+      "input": 0.3,
+      "output": 1.2,
+      "cached_input": 0.06
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "minimax/minimax-m3",
+      "variant": "",
+      "input": 0.3,
+      "output": 1.2,
+      "cached_input": 0.06
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/codestral-2508",
+      "variant": "",
+      "input": 0.3,
+      "output": 0.9,
+      "cached_input": 0.03
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/devstral-2512",
+      "variant": "",
+      "input": 0.4,
+      "output": 2
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/ministral-14b-2512",
+      "variant": "",
+      "input": 0.2,
+      "output": 0.2
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/ministral-3b-2512",
+      "variant": "",
+      "input": 0.1,
+      "output": 0.1
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/ministral-8b-2512",
+      "variant": "",
+      "input": 0.15,
+      "output": 0.15
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mistral-7b-instruct",
+      "variant": "",
+      "input": 0.13,
+      "output": 0.13
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mistral-large",
+      "variant": "",
+      "input": 2,
+      "output": 6
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mistral-large-2407",
+      "variant": "",
+      "input": 2,
+      "output": 6,
+      "cached_input": 0.2
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mistral-large-2512",
+      "variant": "",
+      "input": 0.5,
+      "output": 1.5
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mistral-medium-3",
+      "variant": "",
+      "input": 0.4,
+      "output": 2,
+      "cached_input": 0.04
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mistral-medium-3-5",
+      "variant": "",
+      "input": 1.5,
+      "output": 7.5
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mistral-medium-3.1",
+      "variant": "",
+      "input": 0.4,
+      "output": 2,
+      "cached_input": 0.04
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mistral-nemo",
+      "variant": "",
+      "input": 0.019,
+      "output": 0.03
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mistral-saba",
+      "variant": "",
+      "input": 0.2,
+      "output": 0.6,
+      "cached_input": 0.02
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mistral-small-24b-instruct-2501",
+      "variant": "",
+      "input": 0.05,
+      "output": 0.08
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mistral-small-2603",
+      "variant": "",
+      "input": 0.15,
+      "output": 0.6,
+      "cached_input": 0.015
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mistral-small-3.1-24b-instruct",
+      "variant": "",
+      "input": 0.351,
+      "output": 0.555
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mistral-small-3.2-24b-instruct",
+      "variant": "",
+      "input": 0.075,
+      "output": 0.2
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/mixtral-8x22b-instruct",
+      "variant": "",
+      "input": 2,
+      "output": 6
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "mistralai/voxtral-small-24b-2507",
+      "variant": "",
+      "input": 0.1,
+      "output": 0.3,
+      "cached_input": 0.01
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "moonshotai/kimi-k2",
+      "variant": "",
+      "input": 0.57,
+      "output": 2.3
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "moonshotai/kimi-k2-0905",
+      "variant": "",
+      "input": 0.6,
+      "output": 2.5
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "moonshotai/kimi-k2-thinking",
+      "variant": "",
+      "input": 0.6,
+      "output": 2.5,
+      "cached_input": 0.15
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "moonshotai/kimi-k2.5",
+      "variant": "",
+      "input": 0.45,
+      "output": 2.25,
+      "cached_input": 0.07
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "moonshotai/kimi-k2.6",
+      "variant": "",
+      "input": 0.95,
+      "output": 4,
+      "cached_input": 0.16
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "moonshotai/kimi-k2.7-code",
+      "variant": "",
+      "input": 0.66,
+      "output": 3.4,
+      "cached_input": 0.18
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "moonshotai/kimi-k3",
+      "variant": "",
+      "input": 3,
+      "output": 15,
+      "cached_input": 0.3
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "nvidia/nemotron-3-nano-30b-a3b",
+      "variant": "",
+      "input": 0.05,
+      "output": 0.2,
+      "cached_input": 0.03
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "nvidia/nemotron-3-super-120b-a12b",
+      "variant": "",
+      "input": 0.085,
+      "output": 0.4
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "nvidia/nemotron-3-ultra-550b-a55b",
+      "variant": "",
+      "input": 0.625,
+      "output": 3.125,
+      "cached_input": 0.1875
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "nvidia/nemotron-3.5-content-safety",
+      "variant": "",
+      "input": 0.2,
+      "output": 0.2
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "nvidia/nemotron-3.5-lightning",
+      "variant": "",
+      "input": 0.08,
+      "output": 0.2
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-3.5-turbo",
+      "variant": "",
+      "input": 0.5,
+      "output": 1.5
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-3.5-turbo-16k",
+      "variant": "",
+      "input": 3,
+      "output": 4
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-3.5-turbo-instruct",
+      "variant": "",
+      "input": 1.5,
+      "output": 2
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-4",
+      "variant": "",
+      "input": 30,
+      "output": 60
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-4-turbo",
+      "variant": "",
+      "input": 10,
+      "output": 30
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-4-turbo-preview",
+      "variant": "",
+      "input": 10,
+      "output": 30
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-4.1",
+      "variant": "",
+      "input": 2,
+      "output": 8,
+      "cached_input": 0.5
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-4.1-mini",
+      "variant": "",
+      "input": 0.4,
+      "output": 1.6,
+      "cached_input": 0.1
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-4.1-nano",
+      "variant": "",
+      "input": 0.1,
+      "output": 0.4,
+      "cached_input": 0.025
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-4o",
+      "variant": "",
+      "input": 2.5,
+      "output": 10,
+      "cached_input": 1.25
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-4o-2024-05-13",
+      "variant": "",
+      "input": 5,
+      "output": 15
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-4o-2024-08-06",
+      "variant": "",
+      "input": 2.5,
+      "output": 10,
+      "cached_input": 1.25
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-4o-2024-11-20",
+      "variant": "",
+      "input": 2.5,
+      "output": 10,
+      "cached_input": 1.25
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-4o-mini",
+      "variant": "",
+      "input": 0.15,
+      "output": 0.6,
+      "cached_input": 0.075
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-4o-mini-2024-07-18",
+      "variant": "",
+      "input": 0.15,
+      "output": 0.6,
+      "cached_input": 0.075
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5",
+      "variant": "",
+      "input": 1.25,
+      "output": 10,
+      "cached_input": 0.125
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5-chat",
+      "variant": "",
+      "input": 1.25,
+      "output": 10,
+      "cached_input": 0.125
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5-codex",
+      "variant": "",
+      "input": 1.25,
+      "output": 10,
+      "cached_input": 0.125
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5-mini",
+      "variant": "",
+      "input": 0.25,
+      "output": 2,
+      "cached_input": 0.025
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5-nano",
+      "variant": "",
+      "input": 0.05,
+      "output": 0.4,
+      "cached_input": 0.005
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5-pro",
+      "variant": "",
+      "input": 15,
+      "output": 120
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.1",
+      "variant": "",
+      "input": 1.25,
+      "output": 10,
+      "cached_input": 0.125
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.1-codex",
+      "variant": "",
+      "input": 1.25,
+      "output": 10,
+      "cached_input": 0.13
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.1-codex-max",
       "variant": "",
       "input": 1.25,
       "output": 10,
@@ -10183,229 +11304,779 @@
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-5-chat",
+      "model": "openai/gpt-5.1-codex-mini",
+      "variant": "",
+      "input": 0.25,
+      "output": 2,
+      "cached_input": 0.03
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.2",
+      "variant": "",
+      "input": 1.75,
+      "output": 14,
+      "cached_input": 0.175
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.2-chat",
+      "variant": "",
+      "input": 1.75,
+      "output": 14,
+      "cached_input": 0.175
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.2-codex",
+      "variant": "",
+      "input": 1.75,
+      "output": 14,
+      "cached_input": 0.175
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.2-pro",
+      "variant": "",
+      "input": 21,
+      "output": 168
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.3-codex",
+      "variant": "",
+      "input": 1.75,
+      "output": 14,
+      "cached_input": 0.175
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.4",
+      "variant": "",
+      "input": 2.5,
+      "output": 15,
+      "cached_input": 0.25
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.4-mini",
+      "variant": "",
+      "input": 0.75,
+      "output": 4.5,
+      "cached_input": 0.075
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.4-nano",
+      "variant": "",
+      "input": 0.2,
+      "output": 1.25,
+      "cached_input": 0.02
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.4-pro",
+      "variant": "",
+      "input": 30,
+      "output": 180
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.4-pro",
+      "variant": "above_272k_tokens",
+      "input": 60,
+      "output": 270
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.5",
+      "variant": "",
+      "input": 5,
+      "output": 30,
+      "cached_input": 0.5
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.5-pro",
+      "variant": "",
+      "input": 30,
+      "output": 180
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.5-pro",
+      "variant": "above_272k_tokens",
+      "input": 60,
+      "output": 270
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.6-luna",
+      "variant": "",
+      "input": 0.2,
+      "output": 1.2,
+      "cached_input": 0.02
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-5.6-terra",
+      "variant": "",
+      "input": 2,
+      "output": 12,
+      "cached_input": 0.2
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-6-astra",
+      "variant": "",
+      "input": 10,
+      "output": 50,
+      "cached_input": 1
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-6-astra",
+      "variant": "above_272k_tokens",
+      "input": 20,
+      "output": 75,
+      "cached_input": 2
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-audio",
+      "variant": "",
+      "input": 2.5,
+      "output": 10
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-audio-mini",
+      "variant": "",
+      "input": 0.6,
+      "output": 2.4
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-oss-120b",
+      "variant": "",
+      "input": 0.037,
+      "output": 0.17
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-oss-20b",
+      "variant": "",
+      "input": 0.03,
+      "output": 0.13
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/gpt-oss-safeguard-20b",
+      "variant": "",
+      "input": 0.075,
+      "output": 0.3,
+      "cached_input": 0.0375
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/o1",
+      "variant": "",
+      "input": 15,
+      "output": 60,
+      "cached_input": 7.5
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/o1-pro",
+      "variant": "",
+      "input": 150,
+      "output": 600
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/o3",
+      "variant": "",
+      "input": 2,
+      "output": 8,
+      "cached_input": 0.5
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/o3-mini",
+      "variant": "",
+      "input": 1.1,
+      "output": 4.4,
+      "cached_input": 0.55
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/o3-mini-high",
+      "variant": "",
+      "input": 1.1,
+      "output": 4.4,
+      "cached_input": 0.55
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/o3-pro",
       "variant": "",
-      "input": 1.25,
-      "output": 10,
-      "cached_input": 0.125
+      "input": 20,
+      "output": 80
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/o4-mini",
+      "variant": "",
+      "input": 1.1,
+      "output": 4.4,
+      "cached_input": 0.275
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "openai/o4-mini-high",
+      "variant": "",
+      "input": 1.1,
+      "output": 4.4,
+      "cached_input": 0.275
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "poolside/laguna-s-2.1",
+      "variant": "",
+      "input": 0.09,
+      "output": 0.18,
+      "cached_input": 0.009
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "poolside/laguna-xs-2.1",
+      "variant": "",
+      "input": 0.06,
+      "output": 0.12,
+      "cached_input": 0.03
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen-2.5-72b-instruct",
+      "variant": "",
+      "input": 0.36,
+      "output": 0.4
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen-2.5-7b-instruct",
+      "variant": "",
+      "input": 0.1,
+      "output": 0.2
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen-2.5-coder-32b-instruct",
+      "variant": "",
+      "input": 0.66,
+      "output": 1
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen-plus",
+      "variant": "",
+      "input": 0.26,
+      "output": 0.78,
+      "cached_input": 0.052
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen-plus",
+      "variant": "above_256k_tokens",
+      "input": 0.78,
+      "output": 2.34,
+      "cached_input": 0.156
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen-plus-2025-07-28",
+      "variant": "",
+      "input": 0.26,
+      "output": 0.78
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen-plus-2025-07-28",
+      "variant": "above_256k_tokens",
+      "input": 0.78,
+      "output": 2.34
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen-vl-plus",
+      "variant": "",
+      "input": 0.21,
+      "output": 0.63
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen2.5-vl-72b-instruct",
+      "variant": "",
+      "input": 0.8,
+      "output": 1,
+      "cached_input": 0.4
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-14b",
+      "variant": "",
+      "input": 0.12,
+      "output": 0.24
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-235b-a22b",
+      "variant": "",
+      "input": 0.455,
+      "output": 1.82
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-235b-a22b-2507",
+      "variant": "",
+      "input": 0.0875,
+      "output": 0.35
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-235b-a22b-thinking-2507",
+      "variant": "",
+      "input": 0.23,
+      "output": 2.3
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-30b-a3b",
+      "variant": "",
+      "input": 0.12,
+      "output": 0.5
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-30b-a3b-instruct-2507",
+      "variant": "",
+      "input": 0.04815,
+      "output": 0.19305
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-30b-a3b-thinking-2507",
+      "variant": "",
+      "input": 0.2,
+      "output": 2.4
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-32b",
+      "variant": "",
+      "input": 0.08,
+      "output": 0.28
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-8b",
+      "variant": "",
+      "input": 0.117,
+      "output": 0.455
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-coder",
+      "variant": "",
+      "input": 0.3,
+      "output": 1
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-coder-30b-a3b-instruct",
+      "variant": "",
+      "input": 0.07,
+      "output": 0.28
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-coder-flash",
+      "variant": "",
+      "input": 0.195,
+      "output": 0.975,
+      "cached_input": 0.039
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-coder-flash",
+      "variant": "above_128k_tokens",
+      "input": 0.52,
+      "output": 2.6,
+      "cached_input": 0.104
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-coder-next",
+      "variant": "",
+      "input": 0.12,
+      "output": 0.8,
+      "cached_input": 0.07
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-coder-plus",
+      "variant": "",
+      "input": 0.65,
+      "output": 3.25
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-max",
+      "variant": "",
+      "input": 0.78,
+      "output": 3.9,
+      "cached_input": 0.156
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-max",
+      "variant": "above_128k_tokens",
+      "input": 1.95,
+      "output": 9.75,
+      "cached_input": 0.39
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-max-thinking",
+      "variant": "",
+      "input": 0.78,
+      "output": 3.9
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-max-thinking",
+      "variant": "above_128k_tokens",
+      "input": 1.95,
+      "output": 9.75
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-next-80b-a3b-instruct",
+      "variant": "",
+      "input": 0.1,
+      "output": 1.1,
+      "cached_input": 0.07
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-next-80b-a3b-thinking",
+      "variant": "",
+      "input": 0.15,
+      "output": 1.2
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-vl-235b-a22b-instruct",
+      "variant": "",
+      "input": 0.21,
+      "output": 1.9,
+      "cached_input": 0.1
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-vl-235b-a22b-thinking",
+      "variant": "",
+      "input": 0.4,
+      "output": 4
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-vl-30b-a3b-instruct",
+      "variant": "",
+      "input": 0.15,
+      "output": 0.6
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-vl-30b-a3b-thinking",
+      "variant": "",
+      "input": 0.2,
+      "output": 2.4
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-vl-32b-instruct",
+      "variant": "",
+      "input": 0.104,
+      "output": 0.416
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-vl-8b-instruct",
+      "variant": "",
+      "input": 0.117,
+      "output": 0.455
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3-vl-8b-thinking",
+      "variant": "",
+      "input": 0.18,
+      "output": 2.1
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3.5-122b-a10b",
+      "variant": "",
+      "input": 0.29,
+      "output": 2.4
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3.5-27b",
+      "variant": "",
+      "input": 0.195,
+      "output": 1.56
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3.5-35b-a3b",
+      "variant": "",
+      "input": 0.25,
+      "output": 1.25
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-5-codex",
+      "model": "qwen/qwen3.5-397b-a17b",
       "variant": "",
-      "input": 1.25,
-      "output": 10,
-      "cached_input": 0.125
+      "input": 0.55,
+      "output": 3.5
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-5-mini",
+      "model": "qwen/qwen3.5-9b",
       "variant": "",
-      "input": 0.25,
-      "output": 2,
-      "cached_input": 0.025
+      "input": 0.1,
+      "output": 0.15
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-5-nano",
+      "model": "qwen/qwen3.5-flash-02-23",
       "variant": "",
-      "input": 0.05,
-      "output": 0.4,
-      "cached_input": 0.005
+      "input": 0.065,
+      "output": 0.26
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-5.1-codex-max",
+      "model": "qwen/qwen3.5-plus-02-15",
       "variant": "",
-      "input": 1.25,
-      "output": 10,
-      "cached_input": 0.125
+      "input": 0.26,
+      "output": 1.56
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-5.2",
-      "variant": "",
-      "input": 1.75,
-      "output": 14,
-      "cached_input": 0.175
+      "model": "qwen/qwen3.5-plus-02-15",
+      "variant": "above_256k_tokens",
+      "input": 0.5,
+      "output": 3
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-5.2-chat",
+      "model": "qwen/qwen3.5-plus-20260420",
       "variant": "",
-      "input": 1.75,
-      "output": 14,
-      "cached_input": 0.175
+      "input": 0.3,
+      "output": 1.8
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-5.2-codex",
-      "variant": "",
-      "input": 1.75,
-      "output": 14,
-      "cached_input": 0.175
+      "model": "qwen/qwen3.5-plus-20260420",
+      "variant": "above_256k_tokens",
+      "input": 0.375,
+      "output": 2.25
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-5.2-pro",
+      "model": "qwen/qwen3.6-27b",
       "variant": "",
-      "input": 21,
-      "output": 168
+      "input": 0.3,
+      "output": 2,
+      "cached_input": 0.03
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-oss-120b",
+      "model": "qwen/qwen3.6-35b-a3b",
       "variant": "",
-      "input": 0.18,
-      "output": 0.8
+      "input": 0.1,
+      "output": 0.9,
+      "cached_input": 0.05
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/gpt-oss-20b",
+      "model": "qwen/qwen3.6-flash",
       "variant": "",
-      "input": 0.02,
-      "output": 0.1
+      "input": 0.1875,
+      "output": 1.125
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/o1",
-      "variant": "",
-      "input": 15,
-      "output": 60,
-      "cached_input": 7.5
+      "model": "qwen/qwen3.6-flash",
+      "variant": "above_256k_tokens",
+      "input": 0.75,
+      "output": 3
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/o3-mini",
+      "model": "qwen/qwen3.6-max-preview",
       "variant": "",
-      "input": 1.1,
-      "output": 4.4
+      "input": 1.027,
+      "output": 6.162
     },
     {
       "provider": "OpenRouter",
-      "model": "openai/o3-mini-high",
-      "variant": "",
-      "input": 1.1,
-      "output": 4.4
+      "model": "qwen/qwen3.6-max-preview",
+      "variant": "above_128k_tokens",
+      "input": 1.58,
+      "output": 9.48
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen-2.5-coder-32b-instruct",
+      "model": "qwen/qwen3.6-plus",
       "variant": "",
-      "input": 0.18,
-      "output": 0.18
+      "input": 0.325,
+      "output": 1.95
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen-vl-plus",
+      "model": "qwen/qwen3.7-flash",
       "variant": "",
-      "input": 0.21,
-      "output": 0.63
+      "input": 0.03,
+      "output": 0.13,
+      "cached_input": 0.006
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen3-235b-a22b-2507",
+      "model": "qwen/qwen3.7-flash",
+      "variant": "above_256k_tokens",
+      "input": 0.2,
+      "output": 0.8,
+      "cached_input": 0.04
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "qwen/qwen3.7-max",
       "variant": "",
-      "input": 0.071,
-      "output": 0.1
+      "input": 1.475,
+      "output": 4.425,
+      "cached_input": 0.295
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen3-235b-a22b-thinking-2507",
+      "model": "qwen/qwen3.7-plus",
       "variant": "",
-      "input": 0.11,
-      "output": 0.6
+      "input": 0.32,
+      "output": 1.28,
+      "cached_input": 0.064
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen3-coder",
+      "model": "qwen/qwen3.8-2.4t-a95b",
       "variant": "",
-      "input": 0.22,
-      "output": 0.95
+      "input": 2,
+      "output": 6,
+      "cached_input": 0.25
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen3-coder-plus",
+      "model": "qwen/qwen3.8-27b",
       "variant": "",
-      "input": 1,
-      "output": 5
+      "input": 0.42,
+      "output": 3,
+      "cached_input": 0.085
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen3.5-122b-a10b",
+      "model": "qwen/qwen3.8-flash",
       "variant": "",
-      "input": 0.4,
-      "output": 2
+      "input": 0.15,
+      "output": 0.47,
+      "cached_input": 0.016
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen3.5-27b",
+      "model": "qwen/qwen3.8-max",
       "variant": "",
-      "input": 0.3,
-      "output": 2.4
+      "input": 2,
+      "output": 6,
+      "cached_input": 0.25
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen3.5-35b-a3b",
+      "model": "switchpoint/router",
       "variant": "",
-      "input": 0.25,
-      "output": 2
+      "input": 0.85,
+      "output": 3.4
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen3.5-397b-a17b",
+      "model": "undi95/remm-slerp-l2-13b",
       "variant": "",
-      "input": 0.6,
-      "output": 3.6
+      "input": 0.45,
+      "output": 0.65
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen3.5-flash-02-23",
+      "model": "x-ai/grok-4",
       "variant": "",
-      "input": 0.1,
-      "output": 0.4
+      "input": 3,
+      "output": 15
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen3.5-plus-02-15",
+      "model": "x-ai/grok-4.20",
       "variant": "",
-      "input": 0.4,
-      "output": 2.4
+      "input": 1.25,
+      "output": 2.5,
+      "cached_input": 0.2
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen3.5-plus-02-15",
-      "variant": "above_256k_tokens",
-      "input": 0.5,
-      "output": 3
+      "model": "x-ai/grok-4.20-multi-agent",
+      "variant": "",
+      "input": 1.25,
+      "output": 2.5,
+      "cached_input": 0.2
     },
     {
       "provider": "OpenRouter",
-      "model": "qwen/qwen3.6-plus",
+      "model": "x-ai/grok-4.3",
       "variant": "",
-      "input": 0.325,
-      "output": 1.95
+      "input": 1.25,
+      "output": 2.5,
+      "cached_input": 0.2
     },
     {
       "provider": "OpenRouter",
-      "model": "switchpoint/router",
+      "model": "x-ai/grok-4.5",
       "variant": "",
-      "input": 0.85,
-      "output": 3.4
+      "input": 2,
+      "output": 6,
+      "cached_input": 0.3
     },
     {
       "provider": "OpenRouter",
-      "model": "undi95/remm-slerp-l2-13b",
+      "model": "x-ai/grok-4.6",
       "variant": "",
-      "input": 1.875,
-      "output": 1.875
+      "input": 2,
+      "output": 6,
+      "cached_input": 0.5
     },
     {
       "provider": "OpenRouter",
-      "model": "x-ai/grok-4",
+      "model": "x-ai/grok-build-0.1",
       "variant": "",
-      "input": 3,
-      "output": 15
+      "input": 1,
+      "output": 2,
+      "cached_input": 0.2
     },
     {
       "provider": "OpenRouter",
@@ -10419,24 +12090,48 @@
       "provider": "OpenRouter",
       "model": "xiaomi/mimo-v2.5",
       "variant": "",
-      "input": 0.4,
-      "output": 2,
-      "cached_input": 0.08
+      "input": 0.14,
+      "output": 0.28,
+      "cached_input": 0.0028
     },
     {
       "provider": "OpenRouter",
       "model": "xiaomi/mimo-v2.5-pro",
       "variant": "",
-      "input": 1,
-      "output": 3,
-      "cached_input": 0.2
+      "input": 0.435,
+      "output": 0.87,
+      "cached_input": 0.0036
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "z-ai/glm-4.5",
+      "variant": "",
+      "input": 0.6,
+      "output": 2.2,
+      "cached_input": 0.11
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "z-ai/glm-4.5-air",
+      "variant": "",
+      "input": 0.13,
+      "output": 0.85,
+      "cached_input": 0.025
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "z-ai/glm-4.5v",
+      "variant": "",
+      "input": 0.6,
+      "output": 1.8,
+      "cached_input": 0.11
     },
     {
       "provider": "OpenRouter",
       "model": "z-ai/glm-4.6",
       "variant": "",
-      "input": 0.4,
-      "output": 1.75
+      "input": 0.55,
+      "output": 2.2
     },
     {
       "provider": "OpenRouter",
@@ -10445,36 +12140,84 @@
       "input": 0.45,
       "output": 1.9
     },
+    {
+      "provider": "OpenRouter",
+      "model": "z-ai/glm-4.6v",
+      "variant": "",
+      "input": 0.3,
+      "output": 0.9,
+      "cached_input": 0.055
+    },
     {
       "provider": "OpenRouter",
       "model": "z-ai/glm-4.7",
       "variant": "",
       "input": 0.4,
-      "output": 1.5,
-      "cached_input": 0
+      "output": 1.75,
+      "cached_input": 0.08
     },
     {
       "provider": "OpenRouter",
       "model": "z-ai/glm-4.7-flash",
       "variant": "",
-      "input": 0.07,
+      "input": 0.06,
       "output": 0.4,
-      "cached_input": 0
+      "cached_input": 0.01
     },
     {
       "provider": "OpenRouter",
       "model": "z-ai/glm-5",
       "variant": "",
-      "input": 0.8,
-      "output": 2.56
+      "input": 0.6,
+      "output": 1.92
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "z-ai/glm-5-turbo",
+      "variant": "",
+      "input": 1.2,
+      "output": 4,
+      "cached_input": 0.24
     },
     {
       "provider": "OpenRouter",
       "model": "z-ai/glm-5.1",
       "variant": "",
-      "input": 1.05,
-      "output": 3.5,
-      "cached_input": 0.525
+      "input": 0.966,
+      "output": 3.036,
+      "cached_input": 0.1794
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "z-ai/glm-5.2",
+      "variant": "",
+      "input": 0.966,
+      "output": 3.036,
+      "cached_input": 0.1932
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "z-ai/glm-5.3",
+      "variant": "",
+      "input": 1.4,
+      "output": 4.4,
+      "cached_input": 0.14
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "z-ai/glm-5.3-flash",
+      "variant": "",
+      "input": 0.075,
+      "output": 0.25,
+      "cached_input": 0.015
+    },
+    {
+      "provider": "OpenRouter",
+      "model": "z-ai/glm-5v-turbo",
+      "variant": "",
+      "input": 1.2,
+      "output": 4,
+      "cached_input": 0.24
     },
     {
       "provider": "Posit",
diff --git a/chatlas/types/anthropic/_submit.py b/chatlas/types/anthropic/_submit.py
index af13d8b7..b4361c5d 100644
--- a/chatlas/types/anthropic/_submit.py
+++ b/chatlas/types/anthropic/_submit.py
@@ -126,6 +126,7 @@ class SubmitInputArgs(TypedDict, total=False):
         anthropic.Omit,
     ]
     user_profile_id: str | anthropic.Omit
+    workspace_id: str | anthropic.Omit
     extra_headers: Optional[Mapping[str, Union[str, anthropic.Omit]]]
     extra_query: Optional[Mapping[str, object]]
     extra_body: object | None

From 2d715c879f191beb381153631dc5af1edd455025 Mon Sep 17 00:00:00 2001
From: Carson <cpsievert1@gmail.com>
Date: Tue, 8 Sep 2026 11:25:25 -0500
Subject: [PATCH 4/7] chore: silence pyright reportPrivateImportUsage for
 pillow_heif

CI's pyright resolves pillow_heif's stubs and flags register_heif_opener
as a private re-export; this failure is present on main as well.
---
 chatlas/_content_image.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/chatlas/_content_image.py b/chatlas/_content_image.py
index 70a68a43..351b3797 100644
--- a/chatlas/_content_image.py
+++ b/chatlas/_content_image.py
@@ -156,7 +156,7 @@ def content_image_file(
                 "Install it with `pip install pillow-heif`, or pass "
                 "`resize='none'` to send the original bytes without resizing."
             )
-        pillow_heif.register_heif_opener()
+        pillow_heif.register_heif_opener()  # pyright: ignore[reportPrivateImportUsage]
 
     if resize == "none":
         with open(path, "rb") as image_file:

From ba9fc1a29c9bf55d96422dac42cd78583dc2538d Mon Sep 17 00:00:00 2001
From: Carson <cpsievert1@gmail.com>
Date: Tue, 8 Sep 2026 11:51:04 -0500
Subject: [PATCH 5/7] fix(posit): close old provider and preserve name on
 family switch

Addresses Copilot review feedback on #430:
- Chat.model setter now closes the previous provider's HTTP clients
  when set_model() swaps in a new provider instance (async resources
  are closed via close_async() when an event loop is running).
- Posit set_model() now carries over a custom provider name across
  Anthropic/OpenAI flavor switches.
---
 chatlas/_chat.py             | 16 +++++++++++++-
 chatlas/_provider_posit.py   |  2 ++
 tests/test_provider_posit.py | 41 ++++++++++++++++++++++++++++++++++++
 3 files changed, 58 insertions(+), 1 deletion(-)

diff --git a/chatlas/_chat.py b/chatlas/_chat.py
index 9e3b21c8..a483b17b 100644
--- a/chatlas/_chat.py
+++ b/chatlas/_chat.py
@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import asyncio
 import copy
 import inspect
 import os
@@ -549,7 +550,20 @@ def model(self) -> str:
 
     @model.setter
     def model(self, value: str):
-        self.provider = self.provider.set_model(value)
+        old_provider = self.provider
+        new_provider = old_provider.set_model(value)
+        if new_provider is not old_provider:
+            # A provider swap (e.g., ChatPosit switching API flavors)
+            # abandons the old provider, so release its HTTP clients.
+            # Best-effort for async resources: schedule close_async()
+            # when an event loop is running.
+            try:
+                loop = asyncio.get_running_loop()
+            except RuntimeError:
+                old_provider.close()
+            else:
+                loop.create_task(old_provider.close_async())
+        self.provider = new_provider
 
     @property
     def conversation_id(self) -> str | None:
diff --git a/chatlas/_provider_posit.py b/chatlas/_provider_posit.py
index 20cc5733..efaa4195 100644
--- a/chatlas/_provider_posit.py
+++ b/chatlas/_provider_posit.py
@@ -318,6 +318,7 @@ def set_model(self, value: str) -> "PositAnthropicProvider | PositOpenAIProvider
             model=value,
             credentials=self._credentials,
             cache=self._cache,
+            name=self.name,
         )
 
     def list_models(self) -> list[ModelInfo]:
@@ -371,6 +372,7 @@ def set_model(self, value: str) -> "PositAnthropicProvider | PositOpenAIProvider
                 model=value,
                 credentials=self._credentials,
                 cache=self._cache,
+                name=self.name,
             )
         self._model = value
         return self
diff --git a/tests/test_provider_posit.py b/tests/test_provider_posit.py
index 6d71fac5..73753c7c 100644
--- a/tests/test_provider_posit.py
+++ b/tests/test_provider_posit.py
@@ -612,3 +612,44 @@ def test_chat_posit_set_model_openai_start_honors_cache_on_switch():
 
     assert isinstance(chat.provider, PositAnthropicProvider)
     assert chat.provider._cache == "none"
+
+
+def test_chat_posit_set_model_preserves_custom_name_across_family_switches():
+    provider = PositAnthropicProvider(
+        base_url="https://gateway.posit.ai",
+        model="claude-sonnet-4-6",
+        credentials=lambda: "test-token",
+        name="Custom",
+    )
+
+    openai_provider = provider.set_model("qwen3-8b")
+    assert isinstance(openai_provider, PositOpenAIProvider)
+    assert openai_provider.name == "Custom"
+
+    anthropic_provider = openai_provider.set_model("claude-sonnet-4-6")
+    assert isinstance(anthropic_provider, PositAnthropicProvider)
+    assert anthropic_provider.name == "Custom"
+
+
+def test_chat_posit_set_model_closes_old_provider_on_family_switch():
+    chat = ChatPosit(model="claude-sonnet-4-6", credentials=lambda: "test-token")
+    old_provider = chat.provider
+
+    chat.model = "qwen3-8b"
+
+    assert old_provider._client.is_closed()
+
+    chat = ChatPosit(model="qwen3-8b", credentials=lambda: "test-token")
+    old_provider = chat.provider
+
+    chat.model = "claude-sonnet-4-6"
+
+    assert old_provider._client.is_closed()
+
+
+def test_chat_posit_set_model_within_family_does_not_close_provider():
+    chat = ChatPosit(model="claude-sonnet-4-6", credentials=lambda: "test-token")
+
+    chat.model = "claude-opus-4-1"
+
+    assert not chat.provider._client.is_closed()

From a82b9b834b61cf69d982d377c109af147c12f72e Mon Sep 17 00:00:00 2001
From: Carson <cpsievert1@gmail.com>
Date: Tue, 8 Sep 2026 11:58:23 -0500
Subject: [PATCH 6/7] refactor: simplify model setter to sync-only provider
 close

Async clients can't be closed from sync code (their transport requires
a running event loop), so deterministically closing them from the
Chat.model setter would require fire-and-forget task scheduling with
its own edge cases. The leak scenario (repeated provider swaps after
async usage in a long-lived process) is narrow enough that relying on
GC for async resources is an acceptable, documented tradeoff.
---
 chatlas/_chat.py | 13 ++++---------
 1 file changed, 4 insertions(+), 9 deletions(-)

diff --git a/chatlas/_chat.py b/chatlas/_chat.py
index a483b17b..532dbcab 100644
--- a/chatlas/_chat.py
+++ b/chatlas/_chat.py
@@ -1,6 +1,5 @@
 from __future__ import annotations
 
-import asyncio
 import copy
 import inspect
 import os
@@ -555,14 +554,10 @@ def model(self, value: str):
         if new_provider is not old_provider:
             # A provider swap (e.g., ChatPosit switching API flavors)
             # abandons the old provider, so release its HTTP clients.
-            # Best-effort for async resources: schedule close_async()
-            # when an event loop is running.
-            try:
-                loop = asyncio.get_running_loop()
-            except RuntimeError:
-                old_provider.close()
-            else:
-                loop.create_task(old_provider.close_async())
+            # Only sync resources are closed deterministically: async
+            # clients can't be closed from sync code (their transport
+            # requires a running event loop), so they're left to GC.
+            old_provider.close()
         self.provider = new_provider
 
     @property

From 8179f86314e972184dbc169a6b76d1f93dd3a2c0 Mon Sep 17 00:00:00 2001
From: Carson <cpsievert1@gmail.com>
Date: Tue, 8 Sep 2026 12:02:00 -0500
Subject: [PATCH 7/7] chore: drop explanatory comment from model setter

---
 chatlas/_chat.py | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/chatlas/_chat.py b/chatlas/_chat.py
index 532dbcab..36e592ab 100644
--- a/chatlas/_chat.py
+++ b/chatlas/_chat.py
@@ -552,11 +552,6 @@ def model(self, value: str):
         old_provider = self.provider
         new_provider = old_provider.set_model(value)
         if new_provider is not old_provider:
-            # A provider swap (e.g., ChatPosit switching API flavors)
-            # abandons the old provider, so release its HTTP clients.
-            # Only sync resources are closed deterministically: async
-            # clients can't be closed from sync code (their transport
-            # requires a running event loop), so they're left to GC.
             old_provider.close()
         self.provider = new_provider