fix CI (#1884)

Refactor token parameters to standardize naming across models and generation settings

Omni multimodality (#1880)

HuggingChat 2026 (#1875)

feat: add custom headers support to endpoint inference client

fix: only import node-llama-cpp if needed and skip for huggingchat image

fix: make sure document parser is disabled if not required

New `InferenceClient` endpoint type (#1813)

fix: build local llama (#1805)

feat: allow storing env variable in DB (#1802)

Fix system message handling to preserve user-configured system prompts (#1764)

fix: lint

fix: docker image llama.cpp error catching

Add `toolId` to ToolCall and use it in tool results (#1787)

fix: build llama.cpp locally in dockerfile (#1788)

feat: add a local endpoint type for inference directly from chat-ui (#1778)

fix(preprocessMessages): clarify web search context in final message content

fix(endpoints): fix for tool calling on hf inference with openai endpoint type (#1754)

feat(openai): added support for non-streaming o1 (e.g. Azure) models (#1687)

chores(svelte): migration to svelte 5 (#1685)

fix: bedrock prepareMessages parameters (#1686)

Anthropic Tool Support (#1594)

Add support for Amazon Nova (#1629)

fix: disable caching on OpenAI-compatible endpoint type

fix: single step in cohere endpoint type (#1632)

fix: make single step forcing optional on cohere endpoint type

feat(openai): added support for o1 reasoning models (#1618)

Anthropic PDF Beta Support (#1571)

fix: remove accessToken min length for llama.cpp endpoint

Fix/oai parameters 1552 (#1557)

fix: better websearch quotes

feat: add sources for websearch (#1551)

feat: paste long context as plaintext files (#1549)

[feat] Add support for endpoint web sources (#1544)

fix(endpoint): fallback to hf token for openai endpoint type

[fix] Fix typo in Vertex (#1531)

[vertex] Add PDF/plein texts support (#1520)

feat(endpoints): Add conv ID to headers passed to TGI (#1511)

feat(tools): Basic tool support for OpenAI models (#1447)

Add support for model version in Vertex AI (#1496)

Fix: Cohere Endpoint - Add Support for Tools (#1490)

fix: multimodal on TGI endpoint type (#1491)

fix(endpoints): make bedrock import optional (#1468)

Move Google GenAI API key to env file instead (#1448)

feat: add support for multimodal in Vertex (#1338)

Add support for Anthropic models via AWS Bedrock (#1413)

Dynamic credential resolution for AWS endpoints (#1419)

fix(openai): do not throw "Generation failed" when finish_reason="length" (#1429)

Major bump to sveltekit & node-adapter versions (#1395)

feature(cohere): add clientName param to cohere endpoint