gpt2giga CLI
An ai-forever proxy that serves GigaChat through OpenAI, Anthropic and Gemini APIs so Claude Code, Codex and Gemini CLI can run on it
Broad access
Why this level
- Stores GigaChat credentials and sends the agent's code and session context to GigaChat
- Client key authorization is off by default, and DEV mode exposes /docs and the /logs endpoints
Install
- Works in
- Services
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/gpt2gigaDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Stores GigaChat credentials and sends the agent's code and session context to GigaChat
- Client key authorization is off by default, and DEV mode exposes /docs and the /logs endpoints
This is third-party code. Review the repository files before installing.
What it does
gpt2giga is a FastAPI server that accepts OpenAI, Anthropic and Gemini requests and turns them into GigaChat API calls. For OpenAI it serves /chat/completions, /responses, /embeddings and the model list, for Anthropic /messages and /messages/count_tokens, for Gemini generateContent, streamGenerateContent, countTokens and embedContent. The proxy maps tool calls, structured output, images and SSE streaming, drops unsupported optional fields and keeps the client key separate from GigaChat credentials. It listens on http://localhost:8090 by default, and the /v1 or /v2 path selects the GigaChat v1 or v2 contract.
Who it is for. For developers and teams who need to run Claude Code, Codex, Gemini CLI or their own OpenAI and Anthropic SDK code on GigaChat.
Good fit when
- You want to run Claude Code or Codex on GigaChat-2-Max or GigaChat-3-Ultra
- Existing OpenAI, Anthropic or Gemini SDK code should switch to GigaChat by changing base_url
- You need a shared GigaChat gateway for a team with its own API key and docs closed in PROD mode
Not a fit when
- You need OpenAI, Anthropic or Gemini Files or Batches APIs: the proxy disables them until GigaChat supports them
- You need audio, image generation and editing, fine-tuning or Assistants: these are out of the project's scope
Example request
Start gpt2giga with my GigaChat key and connect Claude Code to it on the GigaChat-2-Max modelLimitations
Requires GigaChat credentials (GIGACHAT_CREDENTIALS) and an account scope such as GIGACHAT_API_PERS. Python 3.10–3.14. Client key authorization is off by default (GPT2GIGA_ENABLE_API_KEY_AUTH=False) and mandatory only in PROD mode, while DEV mode exposes /docs and /logs. The repository has dated verification records for Codex 0.142.1, Claude Code 2.1.187 and Gemini CLI 0.46.0, the Claude Desktop record is marked beta, and the Cursor, Aider, OpenHands and Qwen Code guides have no such record yet.
How to disable. Stop the gpt2giga process, unset ANTHROPIC_BASE_URL and ANTHROPIC_API_KEY or remove the gpt2giga provider from ~/.codex/config.toml, and uninstall the package with uv tool uninstall gpt2giga.
FAQ
Do I need an Anthropic key for Claude Code?
No. Requests go to GigaChat, so any non-empty ANTHROPIC_API_KEY works, or the GPT2GIGA_API_KEY value when proxy authorization is on.
What is the difference between /v1 and /v2?
The /v1 path always selects the GigaChat v1 contract, /v2 selects v2. For Codex with GigaChat built-in tools such as web_search, the guide recommends /v2.
Can each client pass its own GigaChat credentials?
Yes, with GPT2GIGA_PASS_TOKEN=True the proxy accepts them in the Authorization header, though the docs recommend keeping credentials on the server.
Discussion
No questions or reviews yet.
Related
All GigaChat toolsA local proxy exposing GigaChat through OpenAI- and Anthropic-compatible APIs so it can drive Claude Code, aider and other agents
Eight skills for coding against the GigaChat API: choosing a layer, access setup, SDK, LangChain, gpt2giga and Deep Agents
MCP Gateway for GigaChat and YandexGPT
MCP Gateway
An MCP gateway to GigaChat and YandexGPT with automatic failover between them and 152-FZ personal-data anonymization
MCP server for the GigaChat API
GigaChat API
A small Python MCP server for Sber's GigaChat API: authorizes with a Basic key and calls the models directly from an agent, tested on OpenCode under Windows
OpenAI to GigaChat migration skill
migrate-to-gigachat-skill
A Cursor skill that migrates Python, JS or TS projects from OpenAI to GigaChat: detects the stack, picks an adapter, and verifies with tests it still starts
A Home Assistant integration that puts GigaChat, YandexGPT or another LLM behind your smart-home voice assistant
README in short
The README opens by noting GigaChat cannot replace OpenAI or Anthropic without client changes, and lists every supported endpoint of the three APIs. Then it covers installation via uv, pip and Docker Compose, running, the GIGACHAT_* and GPT2GIGA_* variables, OpenAI and Anthropic SDK examples and what is disabled. The GitHub Pages docs cover configuration, deployment behind a reverse proxy and step-by-step integrations with Claude Code, Codex, Gemini CLI, Cursor and other clients.