[cli] Rest API server in local foundry cli-preview-0.10.0

[omiossec]

---

Problem

The documentation Foundry Local REST API Reference is outdated and none of the URI work.

Diagnostics

Looking at the webservice.cc file, there are a new set of URI

`
// Status
impl_->router->route("GET", "/status",
std::make_shared(ctx));

// Shutdown
impl_->router->route("POST", "/shutdown",
std::make_shared(impl_->shutdown_callback));

// Model management
impl_->router->route("GET", "/models/loaded", CreateListLoadedModelsHandler(ctx));
impl_->router->route("GET", "/models/load/{name}", CreateLoadModelHandler(ctx));
impl_->router->route("GET", "/models/unload/{name}", CreateUnloadModelHandler(ctx));

// OpenAI-compatible endpoints
impl_->router->route("GET", "/v1/models", CreateOpenAIListModelsHandler(ctx));
impl_->router->route("GET", "/v1/models/{name}", CreateOpenAIRetrieveModelHandler(ctx));
impl_->router->route("POST", "/v1/chat/completions", CreateChatCompletionsHandler(ctx));
impl_->router->route("POST", "/v1/audio/transcriptions", CreateAudioTranscriptionsHandler(ctx));
impl_->router->route("POST", "/v1/embeddings", CreateEmbeddingsHandler(ctx));
impl_->router->route("POST", "/v1/responses", CreateResponsesHandler(ctx));
impl_->router->route("GET", "/v1/responses", CreateListResponsesHandler(ctx));
impl_->router->route("GET", "/v1/responses/{id}", CreateGetResponseHandler(ctx));
impl_->router->route("DELETE", "/v1/responses/{id}", CreateDeleteResponseHandler(ctx));
impl_->router->route("GET", "/v1/responses/{id}/input_items", CreateGetInputItemsHandler(ctx));
`
Have you an updated documentation?

System

• OS: Microsoft Windows 10.0.26200
• Architecture: Arm64
• CPU: Snapdragon(R) X - X126100 - Qualcomm(R) Oryon(TM) CPU (8 logical cores)
• GPU: Qualcomm Incorporated Qualcomm(R) Adreno(TM) X1-45 GPU (—)
• NPU: Qualcomm Technologies, Inc. Snapdragon(R) X - X126100 - Qualcomm(R) Hexagon(TM) NPU
• RAM: 2.0 GB available / 15.6 GB total
• Disk: 667.7 GB free / 933.0 GB total (C:)
• .NET: .NET 9.0.16

Service

• State: Ready
• Web URLs: http://127.0.0.1:58113
• CLI version: 0.10.0

Models

• Available variants: 47
• Loaded: 0
• Cached: 8

Connectivity

• Local service: Reachable
• Model registry: Reachable

Warnings

• Qualcomm Adreno GPU detected. If acceleration fails, try a CPU model variant or update the GPU driver.

---

This is an auto-generated diagnostic report for the foundry-local-cli GitHub issue tracker, capturing your system specs, service status, and model cache
state.

[redwyre]

Yeah, this breaks everything that tries to use the REST API.

To the devs: If you're going to break everything, I'd consider moving the openai /v1/ endpoints to /openai/v1/, I did a quick check and everything that accepts an OpenAI REST API takes a base URL so it will work fine and not pollute the root. And keeping /foundry/ for the foundry REST API (ideally with /v1/) keeps it clear which is which and keeps it open for future APIs.

It should also be possible to add toggleable redirects from the old URLs to the new URLs which would make it backwards compatible again... but apps like Visual Studio need to fix their base URL, or just allow any base URL for OpenAI compatible.

[omiossec]

It's seems that the change looks like that:

Action	Old path (< 0.10.0)	New path (≥ 0.10.0)
Service status	GET /openai/status	GET /status
Full model catalogue	GET /foundry/list	GET /v1/models
Loaded models	GET /openai/models	GET /models/loaded
Load a model	POST /openai/load/{name}	POST /models/load/{name}
Unload a model	POST /openai/unload/{name}	POST /models/unload/{name}
Download a model	POST /openai/download	(removed — throws error)
Token count	POST /v1/chat/completions/tokenizer/encode/count	(removed — throws error)
Chat completion	POST /v1/chat/completions	POST /v1/chat/completions (unchanged)
Audio transcription	POST /v1/audio/transcriptions	POST /v1/audio/transcriptions (unchanged)