Skip to content

feat: add Inference APIs as an LLM provider - #13293

Open
TaylorHawkes wants to merge 1 commit into
continuedev:mainfrom
TaylorHawkes:add-inferenceapis-provider
Open

TaylorHawkes wants to merge 1 commit into
continuedev:mainfrom
TaylorHawkes:add-inferenceapis-provider

Conversation

@TaylorHawkes

Copy link
Copy Markdown

Description

Adds Inference APIs as an LLM provider, following the same pattern as the Tensorix provider (#11184).

Inference APIs is an OpenAI-compatible host for open-weight models (DeepSeek V4 Flash / V4.1 Flash / V4 Pro, GLM 5.3 / 5.3 Flash, Kimi K2.7 Code, MiniMax M3, GPT-OSS 120B, Qwen3-VL 235B, Llama 3.3 70B) plus embeddings (bge-m3, Qwen3-Embedding-8B) and rerank (Qwen3-Reranker-8B), at https://api.inferenceapis.com/v1. Models are hosted in the United States with zero data retention and billed per token.

Changes

  • core/llm/llms/InferenceAPIs.ts — provider class extending OpenAI (default model zai-org/GLM-5.3)
  • core/llm/llms/index.ts — import and register
  • core/llm/autodetect.ts — templating, images (Qwen3-VL) and parallel-request lists
  • packages/openai-adapters/src/index.ts, types.tsinferenceapis case and zod literal
  • extensions/vscode/config_schema.json — provider enum + description, embeddings and reranker enums
  • gui/src/pages/AddNewModel/configs/providers.ts — Add Model entry
  • docs/customize/model-providers/more/inferenceapis.mdx, docs/docs.json — provider page (chat, autocomplete, embeddings) and nav entry

Notes

  • core/control-plane/schema.ts from the Tensorix diff no longer exists; its successor packages/config-types/src/index.ts does not list Tensorix or the other OpenAI-compatible providers either, so it was left alone, as in feat: add MiniMax as a new LLM provider with M2.7 default #11367.
  • Verified: both JSON files parse; the enum / markdownEnumDescriptions arrays in config_schema.json stay aligned by index; tsc --noEmit is clean in packages/openai-adapters, and in core/ the error count is unchanged from main (pre-existing missing-workspace-package errors). The GUI type check and test suites were not run.
  • Endpoint verified separately: streaming chat, tool calls and embeddings through the OpenAI SDKs, and end-to-end runs of opencode and Codex CLI on the same base URL.
  • I run Inference APIs.

🤖 Generated with Claude Code

https://claude.ai/code/session_01NatZMWWFejKjLafCN9N6kU

@TaylorHawkes
TaylorHawkes requested a review from a team as a code owner September 18, 2026 02:49
@TaylorHawkes
TaylorHawkes requested review from sestinj and removed request for a team September 18, 2026 02:49
@github-actions

github-actions Bot commented Sep 18, 2026

Copy link
Copy Markdown
Contributor

All contributors have signed the CLA ✍️ ✅
Posted by the CLA Assistant Lite bot.

Adds Inference APIs (https://inferenceapis.com), an OpenAI-compatible host for open-weight models, following the same pattern as the Tensorix provider (continuedev#11184).

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NatZMWWFejKjLafCN9N6kU
@TaylorHawkes
TaylorHawkes force-pushed the add-inferenceapis-provider branch from 747be7c to 9f63c23 Compare September 18, 2026 14:12
@TaylorHawkes

Copy link
Copy Markdown
Author

I have read the CLA Document and I hereby sign the CLA

@TaylorHawkes

Copy link
Copy Markdown
Author

I have read the CLA Document and I hereby sign the CLA.

@TaylorHawkes

Copy link
Copy Markdown
Author

I have read the CLA Document and I hereby sign the CLA

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant