# lm-proxy

> OpenAI-compatible HTTP LLM proxy / gateway for multi-provider inference (Google, Anthropic, OpenAI, PyTorch). Lightweight, extensible Python/FastAPI—use as library or standalone service.

- **URL**: https://www.freshcrate.ai/projects/lm-proxy
- **Author**: Nayjest
- **Category**: Infrastructure
- **Latest version**: `v3.2.2` (2026-04-02)
- **License**: MIT
- **Source**: https://github.com/Nayjest/lm-proxy
- **Language**: Python
- **GitHub**: 114 stars, 13 forks
- **Registry**: github
- **Tags**: `ai`, `anthropic`, `api-proxy`, `fastapi`, `google-ai`, `language-models`, `llm`, `llm-api`, `python`

## Description

OpenAI-compatible HTTP LLM proxy / gateway for multi-provider inference (Google, Anthropic, OpenAI, PyTorch). Lightweight, extensible Python/FastAPI—use as library or standalone service.

## Recent releases

| Version | Date | Urgency | Changes |
| --- | --- | --- | --- |
| `v3.2.2` | 2026-04-02 | High | upper-bound constraints on typer and requests dependencies, replacing the compatible-release specifier for requests with an explicit range  ## What's Changed Upper-bound constraints on typer and requests dependencies, replacing the compatible-release specifier for requests with an explicit range   **Full Changelog**: https://github.com/Nayjest/lm-proxy/compare/v3.2.1...v3.2.2 |
| `v3.2.1` | 2026-03-31 | Medium | ## What's Changed - websockets lib updated to v15+   **Full Changelog**: https://github.com/Nayjest/lm-proxy/compare/v3.2.0...v3.2.1 |
| `v3.2.0` | 2026-03-30 | Medium | ## What's Changed - ai-microcore v6 support   **Full Changelog**: https://github.com/Nayjest/lm-proxy/compare/v3.1.0...v3.2.0 |
| `v3.1.0` | 2026-03-25 | Medium | ## What's Changed  ### New: Fallback Strategy  Built-in support for automatic failover across LLM connections. If one provider fails, the next one in the chain is tried — no custom code needed. ```toml [connections.fallback] class = "lm_proxy.strategies.Fallback" connections = ["local.openai/gpt-oss-20b", "openai.gpt-5.2"] ```  - Configurable via list (shorthand) or dict (with per-connection parameter overrides) - Composes with other strategies like load balancing - [Documentation]( |
| `v3.0.2` | 2026-02-19 | Low | ## What's Changed * Fix issues related to bugs in new websockets v15 lib `invalid syntax (http11.py, line 428)` [fix dep. to < v15] https://github.com/Nayjest/lm-proxy/issues/40  ## Non-functional changes * fix coverage badge generation #44  in https://github.com/Nayjest/lm-proxy/pull/45 * Dev-Infrastructure improvements  in https://github.com/Nayjest/lm-proxy/pull/46    **Full Changelog**: https://github.com/Nayjest/lm-proxy/compare/v3.0.1...v3.0.2 |
| `v3.0.1` | 2026-02-10 | Low | ## What's Changed * Fix #41 fix tools usage, forwarding all request parameters supported by OpenAI by @Nayjest in https://github.com/Nayjest/lm-proxy/pull/43 Thanks to @danny70437 for reporting   **Full Changelog**: https://github.com/Nayjest/lm-proxy/compare/v3.0.0...v3.0.1 |
| `v3.0.0` | 2026-02-05 | Low | ## What's Changed * Migrate from ai-microcore v4 to v5 by @Nayjest in https://github.com/Nayjest/lm-proxy/pull/24 (affects configuration format, as lm-proxy directly translates settings to ai-microcore configurations, see [ai-microcore v5 release notes](https://github.com/Nayjest/ai-microcore/releases/tag/v5.0.0)) * HTTP Header Management (Static pre-configured extra-headers + flexible HTTP headers forwarding setup), thanks to @manascb1344 see [HTTP Header Management Article](https://github.co |
| `v2.1.1` | 2025-11-20 | Low | ## What's Changed * Bump version to 2.1.1 and update FastAPI dependency by @NIKITASAGA in https://github.com/Nayjest/lm-proxy/pull/18  ## New Contributors * @NIKITASAGA made their first contribution in https://github.com/Nayjest/lm-proxy/pull/18  **Full Changelog**: https://github.com/Nayjest/lm-proxy/compare/v2.1.0...v2.1.1 |
| `v2.1.0` | 2025-11-02 | Low | ## Added  - **HTTPS Support**: New `ssl_keyfile` and `ssl_certfile` configuration options to enable secure HTTPS connections - **Flexible API Routing**: New `api_path_prefix` option (default: `/v1`) allows customization of endpoint paths. Set to empty string for unversioned paths - **Enhanced Logging**: Support for list indices when extracting target values from `request_context` for improved log detail - **Dynamic Connection Loading**: `config.connections` now accepts strings as Python mod |
| `v2.0.0` | 2025-10-26 | Low | ## What's Changed - Major refactoring - Added component support - `config.check_api_key` renamed to `config.api_key_check` - Added config loaders for **TOML**, **YAML**, **JSON**, and **Python**; custom config loaders can be registered via the `config.loaders` entry point - The `config` module no longer instantiates logger instances — this is now handled in the `bootstrap` process - `LogEntry` has been renamed to `RequestContext` and now includes additional fields such as `model`, `user_in |

## Citation

- HTML: https://www.freshcrate.ai/projects/lm-proxy
- Markdown: https://www.freshcrate.ai/projects/lm-proxy.md
- Dependencies JSON: https://www.freshcrate.ai/api/projects/lm-proxy/deps

_Generated by freshcrate.ai. Indexes github releases for AI-agent ecosystem packages._