raullenchai

raullenchai/Rapid-MLX

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

3.2K stars 376 forks 3.2K watchers Python Apache License 2.0

apple-silicon claude-code cursor deepseek fastapi hacktoberfest inference llm local-llm m1 m2 m3 macos mlx ollama-alternative openai-api python qwen tool-calling

View on GitHub Website

31 Open Issues Need Help Last updated: Jul 7, 2026

Open Issues Need Help

View All on GitHub

Auth hardening follow-ups from #260 and #262 9 days ago

enhancement good first issue security

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: migrate legacy reasoning/tool parsers to token-level OutputRouter 13 days ago

enhancement good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: add Liquid LFM2 support (454K downloads) — needs parser research about 1 month ago

enhancement help wanted difficulty: medium hacktoberfest

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

MLLMScheduler silently ignores frequency_penalty / presence_penalty / repetition_penalty on vision models 2 months ago

bug help wanted

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

model-support: add missing Qwen3.6 MLX aliases (unsloth UD variants + mlx-community 8-bit) 3 months ago

enhancement good first issue model-support

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

Re-cherry-pick upstream security fixes from waybarrios (was #113) 3 months ago

good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

Test Rapid-MLX with more AI clients and report results 3 months ago

documentation help wanted good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

Add logprobs support to chat completions 4 months ago

enhancement help wanted good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

Add request logging middleware 4 months ago

enhancement help wanted good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: migrate Qwen3 reasoning parser to OutputRouter (token-level) 4 months ago

enhancement good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: add GLM-4.6V Flash support (219K downloads, vision model) 4 months ago

enhancement help wanted difficulty: medium hacktoberfest

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: add NVIDIA Nemotron-3 support (178K downloads) — needs parser research 4 months ago

enhancement help wanted difficulty: medium hacktoberfest

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: add Ministral 3B alias 4 months ago

enhancement help wanted good first issue difficulty: easy hacktoberfest

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: add Hermes 4 70B alias (65K downloads) 4 months ago

enhancement help wanted good first issue difficulty: easy hacktoberfest

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: add Qwen3-VL 8B and 30B aliases (vision models) 4 months ago

enhancement help wanted good first issue difficulty: easy hacktoberfest

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: update Devstral alias to v2 (2512 version, 180K downloads) 4 months ago

enhancement help wanted good first issue difficulty: easy hacktoberfest

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: add Qwen3-Coder-30B alias (214K downloads) 4 months ago

enhancement help wanted good first issue difficulty: easy hacktoberfest

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: add Gemma 3n E4B alias (327K downloads) 4 months ago

enhancement help wanted good first issue difficulty: easy hacktoberfest

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: add Kimi K2.5 alias (716K downloads) 4 months ago

enhancement help wanted good first issue difficulty: easy hacktoberfest

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: add Gemma 3 aliases (4B, 1B, 27B) 4 months ago

enhancement help wanted good first issue difficulty: easy hacktoberfest

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: migrate GPT-OSS/Harmony reasoning parser to OutputRouter (token-level) 4 months ago

enhancement good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

feat: migrate DeepSeek R1 reasoning parser to OutputRouter (token-level) 4 months ago

enhancement good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

Add benchmarking script for comparing with Ollama 4 months ago

documentation help wanted good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

Add Dockerfile for easy deployment 4 months ago

enhancement help wanted good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

Support GGUF model loading 4 months ago

enhancement help wanted

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

Add health check endpoint at /health 4 months ago

enhancement help wanted good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

Add /v1/completions token usage tracking 4 months ago

bug help wanted good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

Add --log-level CLI flag 4 months ago

enhancement help wanted good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

Add /v1/embeddings endpoint 4 months ago

enhancement help wanted good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

Document streaming JSON mode limitation 4 months ago

bug help wanted good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling

Add CORS middleware for browser-based clients 5 months ago

enhancement good first issue

raullenchai/Rapid-MLX

3.2K

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Python

#apple-silicon#claude-code#cursor#deepseek#fastapi#hacktoberfest#inference#llm#local-llm#m1#m2#m3#macos#mlx#ollama-alternative#openai-api#python#qwen#tool-calling