Source: https://seerepo.com/compare/hpcaitech/ColossalAI/vs/THUDM/slime
Markdown source: https://seerepo.com/compare/hpcaitech/ColossalAI/vs/THUDM/slime.md
Description: ColossalAI and slime share 8 technologies: FastAPI, Ray, OpenAI, transformers, datasets, safetensors, accelerate, and pillow. Both are written in Python.

# ColossalAI vs slime

ColossalAI and slime share 8 technologies: FastAPI, Ray, OpenAI, transformers, datasets, safetensors, accelerate, and pillow. Both are written in Python.

- [hpcaitech/ColossalAI](https://seerepo.com/r/hpcaitech/ColossalAI): Making large AI models cheaper, faster and more accessible
- [THUDM/slime](https://seerepo.com/r/THUDM/slime): slime is an LLM training framework for RL Scaling.

## What they share and what sets them apart

- In common: FastAPI, Ray, OpenAI, transformers, datasets, safetensors, accelerate, pillow
- Only in ColossalAI: uvicorn, chromadb, Hugging Face, anyio, Gradio, seaborn, matplotlib, Rich, SQLAlchemy, ModelScope, rouge, scikit-learn, sentence-transformers, tiktoken, torch, colossalai, LangChain, triton, peft, diffusers, flash-attn, bitsandbytes, vLLM, einops, galore-torch, loralib, math-verify, sentencepiece, torchaudio, torchvision, jinja2, markdown-it-py, markupsafe, mdurl, pypdf, unstructured, jsonlines, annotated-types, Pydantic, gpustat, psutil, fabric
- Only in slime: blobfile, MCP, openai-agents, e2b, qwen-vl-utils, pylatexenc

## Side by side

| | ColossalAI | slime |
|---|---|---|
| Size | 319,283 lines written, 975 code files | 111,697 lines written, 209 code files |
| Who writes it | 0.3% of its commits signed by AI. With Codex and Claude Code | 3.2% of its commits signed by AI. With Claude Code, Codex, GitHub Copilot, Cursor, and Gemini CLI |
| Activity | Last push November 13, 2026, 3,904 commits, 41,424 stars | Last push November 5, 2026, 1,603 commits, 8,828 stars |
| Tests | 515 test files. 1.1% of its code files have a test that imports them | 170 test files. 65.6% of its code files have a test that imports them |

- Size: Lines written by people and agents, without generated ones; code files, without tests.
- Who writes it: A commit counts as AI's when it carries its signature: a Co-Authored-By line, its author name or the merge of its branch.
- Activity: From GitHub and its history, on the day SeeRepo read it.
- Tests: The files that test the code, and the share of code files a test imports.

## Frequently asked questions

### Which is bigger, ColossalAI or slime?

ColossalAI, with 319,283 lines written; slime has 111,697.

The proof: [hpcaitech/ColossalAI's code at commit 297c8cf](https://github.com/hpcaitech/ColossalAI/tree/297c8cf6d91cb0bff795cba961d3aea1e66fa370)

### Which has more GitHub stars, ColossalAI or slime?

ColossalAI, with 41,424 stars; slime has 8,828.

The proof: [hpcaitech/ColossalAI on GitHub](https://github.com/hpcaitech/ColossalAI)

### Which has more tests, ColossalAI or slime?

ColossalAI, with 515 test files; slime has 170.

The proof: [hpcaitech/ColossalAI's code at commit 297c8cf](https://github.com/hpcaitech/ColossalAI/tree/297c8cf6d91cb0bff795cba961d3aea1e66fa370)

### Which was updated more recently, ColossalAI or slime?

ColossalAI: its last push was on November 13, 2026, and slime's on November 5, 2026.

The proof: [hpcaitech/ColossalAI's commits on GitHub](https://github.com/hpcaitech/ColossalAI/commits)

### Which has more commits signed by AI, ColossalAI or slime?

slime: an AI signed 3.2% of its commits, with Claude Code, Codex, GitHub Copilot, Cursor, and Gemini CLI; in ColossalAI, 0.3%.

The proof: [the commits each tool signed in THUDM/slime](https://seerepo.com/r/THUDM/slime/ai)

## Sources

- hpcaitech/ColossalAI: Read on November 13, 2026, at commit 297c8cf (https://github.com/hpcaitech/ColossalAI/tree/297c8cf6d91cb0bff795cba961d3aea1e66fa370)
- THUDM/slime: Read on November 16, 2026, at commit 79aa41a (https://github.com/THUDM/slime/tree/79aa41a884ab0bf65222c94418f6357d3c31192b)

## What is your repo built with?

Any public repo, with no account: https://seerepo.com/r/owner/repo/built, with the owner and name it has on GitHub.
