Rust Python Accelerators for AI Frameworks
Four Rust-powered drop-in replacements for the Python libraries your AI agents depend on: fast-litellm, fast-langgraph, fast-crewai, and fast-axolotl. One import. Zero config. Up to 737× faster with zero code changes.
Every accelerator is MIT licensed, ships prebuilt wheels for Linux, macOS, and Windows (Python 3.8-3.13), and uses automatic monkey-patching with version detection and silent fallback. No Rust toolchain required.
| Accelerator | Replaces | Headline speedup | Install | Compare |
|---|---|---|---|---|
fast-litellm | LiteLLM | 3.2× faster, 42× less memory | pip install fast-litellm | vs LiteLLM → |
fast-langgraph | LangGraph | 737× faster checkpoints | pip install fast-langgraph | vs LangGraph → |
fast-crewai | CrewAI | 34× faster serialization | pip install fast-crewai | vs CrewAI → |
fast-axolotl | Axolotl | OOM-free training | pip install fast-axolotl | — |
How drop-in acceleration works
# Step 1: Install (no Rust toolchain needed)
$ pip install fast-litellm
# Step 2: Add one import (before litellm)
import fast_litellm
import litellm
# Step 3: There is no step 3
# Your existing code is now 3.2x faster
The import fast_litellm statement triggers automatic monkey-patching that replaces LiteLLM's hot-path implementations (connection pooling, rate limiting, token counting) with Rust equivalents. The library detects the installed version of LiteLLM and patches only compatible code paths, with automatic fallback to the original Python implementation if any incompatibility is detected.
Rust-based acceleration layer for LiteLLM operations. Integrates seamlessly with a single import statement.
$ pip install fast-litellmRust-accelerated library optimizing LangGraph for production workloads with full API compatibility.
$ pip install fast-langgraphUp to 34x faster serialization for CrewAI projects, removing the key bottleneck in multi-agent workflows.
$ pip install fast-crewaiFrequently asked questions
What is a Python accelerator for AI frameworks?
A Python accelerator is a drop-in replacement that makes an existing Python library faster without changing its API. Neul Labs ships four: fast-litellm (3.2× faster connection pooling, 42× less memory), fast-langgraph (737× faster checkpoint serialization), fast-crewai (34× faster serialization), and fast-axolotl (OOM-free large dataset training). All are MIT licensed, ship prebuilt wheels for Linux/macOS/Windows, and integrate with a single import line.
How does fast-litellm compare to LiteLLM?
fast-litellm is a Rust+PyO3 drop-in for LiteLLM. Same Python API. 3.2× faster connection pool acquisition (0.97ms vs 3.1ms at 50 concurrent agents), 42× less memory for high-cardinality rate limiting (0.055MB vs 2.3MB for 10K entries), 1.6-1.7× faster token counting. Install: pip install fast-litellm. Zero code changes — import fast_litellm before litellm. See /compare/fast-litellm-vs-litellm for the full comparison.
How does fast-langgraph make LangGraph faster?
fast-langgraph replaces LangGraph's pure-Python checkpoint serializer with Rust + serde + SIMD, parallelises across Rust threads (bypassing the GIL), and adds batched SQLite writes with WAL mode. 737× faster checkpoint writes (0.1ms vs 73.5ms), 151× faster reads, 46× faster full state-update cycle, 2.8× faster end-to-end 10-step agent execution. Install: pip install fast-langgraph. See /compare/fast-langgraph-vs-langgraph for the full comparison.
Do I need to install Rust to use these accelerators?
No. All four accelerators (fast-litellm, fast-langgraph, fast-crewai, fast-axolotl) ship prebuilt binary wheels for Linux (x86_64, aarch64), macOS (Intel, Apple Silicon), and Windows (x86_64). pip install is all you need. The Rust toolchain is only required if you want to build from source.
Will these accelerators break my existing code?
No. Every accelerator uses automatic monkey-patching with version detection and silent fallback. If the installed version of LiteLLM / LangGraph / CrewAI / Axolotl is incompatible, the library reverts to the original Python implementation with no error. Your tests, configuration, and application code remain unchanged.
What is the best LiteLLM alternative for production?
fast-litellm is the best drop-in replacement for LiteLLM if you need production performance. It preserves the full LiteLLM API, adds Rust-backed connection pooling and rate limiting, and ships as a single pip install. 3.2× faster connections, 42× less memory. MIT licensed. See /compare/fast-litellm-vs-litellm.
What is the best LangGraph optimization for production?
fast-langgraph is the best drop-in optimization for LangGraph if checkpoint overhead is your bottleneck. 737× faster checkpoint serialization, 151× faster deserialization, 2.8× faster end-to-end. MIT licensed. See /compare/fast-langgraph-vs-langgraph.
How do Rust Python accelerators work?
Rust Python accelerators use PyO3 (Rust bindings for Python) to write performance-critical code in Rust and expose it as native Python objects with no serialization cost. The Rust code runs independently of the GIL, enabling true parallelism. The Python API surface is preserved via automatic monkey-patching at import time. See /blog/why-rust-for-ai-infrastructure for the deep dive.