Source: https://seerepo.com/compare/unslothai/unsloth/vs/xorbitsai/inference
Markdown source: https://seerepo.com/compare/unslothai/unsloth/vs/xorbitsai/inference.md
Description: unsloth and inference share 32 technologies: React, FastAPI, Hugging Face, OpenAI, uvicorn, anyio, torch, transformers, peft, accelerate, safetensors, sentence-transformers, gguf, librosa, pillow, Pydantic, psutil, matplotlib, descript-audiotools, einops, julius, sentencepiece, tiktoken, timm, diffusers, loguru, class-variance-authority, Tailwind CSS, tailwind-merge, Radix UI, sonner, and Lucide. Both are written in Python.

# unsloth vs inference

unsloth and inference share 32 technologies: React, FastAPI, Hugging Face, OpenAI, uvicorn, anyio, torch, transformers, peft, accelerate, safetensors, sentence-transformers, gguf, librosa, pillow, Pydantic, psutil, matplotlib, descript-audiotools, einops, julius, sentencepiece, tiktoken, timm, diffusers, loguru, class-variance-authority, Tailwind CSS, tailwind-merge, Radix UI, sonner, and Lucide. Both are written in Python.

- [unslothai/unsloth](https://seerepo.com/r/unslothai/unsloth): Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
- [xorbitsai/inference](https://seerepo.com/r/xorbitsai/inference): Open-source inference platform for language, vision, embedding, reranking, image, audio, and video models. Deploy with multiple engines across heterogeneous hardware, from a laptop to a distributed cluster. Serve models through OpenAI-compatible APIs.

## What they share and what sets them apart

- In common: React, FastAPI, Hugging Face, OpenAI, uvicorn, anyio, torch, transformers, peft, accelerate, safetensors, sentence-transformers, gguf, librosa, pillow, Pydantic, psutil, matplotlib, descript-audiotools, einops, julius, sentencepiece, tiktoken, timm, diffusers, loguru, class-variance-authority, Tailwind CSS, tailwind-merge, Radix UI, sonner, Lucide
- Only in unsloth: sqlite-vec, usearch, fastmcp, Tauri, Tokio, objc2-foundation, glib, objc2-app-kit, gio, TanStack, Vite, dexie, @tauri-apps/api, @assistant-ui/core, Rich, trl, datasets, tokenizers, markdown-it-py, pymupdf, pymupdf4llm, python-docx, av, jinja2, structlog, cairo-rs, gtk, tauri-plugin-dialog, webkit2gtk, webview2-com, gdk, objc2-web-kit, tauri-plugin-window-state, webkit2gtk-sys, tauri-plugin-notification, cryptography, hmac, pyjwt, cut-cross-entropy, descript-audio-codec, einx, jiwer, numkong, openai-whisper, pystoi, scikit-learn, torch-c-dlpack-ext, torch-stoi, triton-kernels, data-designer-engine, snac, kernels, spandrel, markdown2, mdurl, gdk-pixbuf, lxml, log, simplelog, PostCSS, @tailwindcss/vite, postcss-selector-parser, xyflow, motion, @streamdown/code, recharts, cmdk, canvas-confetti, react-day-picker, react-resizable-panels, @dagrejs/dagre, @tauri-apps/plugin-notification, Base UI, hugeicons, @assistant-ui/react, assistant-stream, streamdown, mdast-util-from-markdown, mdast-util-gfm, mdast-util-math, micromark-extension-gfm, micromark-extension-math, unified, katex, remend, js-yaml, mammoth, marked, react-pdf, react-qr-code, decode-named-character-reference, dingbat-to-unicode, micromark-util-types, unpdf, @tauri-apps/plugin-process
- Only in inference: pyarrow, Lightning, AIOHTTP, Next.js, python-jose, bcrypt, en-core-web-sm, misaki, opencv-contrib-python-headless, pynini, transformers-stream-generator, wetextprocessing, torchaudio, torchvision, ModelScope, onnxruntime, x-transformers, qwen-vl-utils, funasr, bitsandbytes, loralib, pyworld, flagembedding, gruut, kokoro, silero-vad, torchdiffeq, vector-quantize-pytorch, vocos, conformer, controlnet-aux, fugashi, g2p-en, langdetect, optimum, tensorizer, eva-decord, imageio-ffmpeg, torchcodec, jsonschema, opentelemetry-instrumentation-httpx, boto3, OpenTelemetry, opentelemetry-exporter-otlp-proto-grpc, opentelemetry-instrumentation-fastapi, next-themes, react-icons, highlight.js, Nunjucks, react-markdown, rehype-katex, rehype-raw, rehype-sanitize, remark-gfm, remark-math

## Side by side

| | unsloth | inference |
|---|---|---|
| Size | 4,065,144 lines written, 3,678 code files | 922,374 lines written, 2,727 code files |
| Who writes it | 1% of its commits signed by AI. With Claude Code, Gemini CLI, Cursor, Codex, and GitHub Copilot | 0.3% of its commits signed by AI. With Gemini CLI, Claude Code, and Cursor |
| Activity | Last push November 12, 2026, 13,457 commits, 79,373 stars | Last push November 14, 2026, 2,791 commits, 9,655 stars |
| Tests | 6,084 test files. 61.5% of its code files have a test that imports them | 676 test files. 19.9% 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, unsloth or inference?

unsloth, with 4,065,144 lines written; inference has 922,374.

The proof: [unslothai/unsloth's code at commit 98836c2](https://github.com/unslothai/unsloth/tree/98836c27662fcce34c60a6e6ca264d5973cb0058)

### Which has more GitHub stars, unsloth or inference?

unsloth, with 79,373 stars; inference has 9,655.

The proof: [unslothai/unsloth on GitHub](https://github.com/unslothai/unsloth)

### Which has more tests, unsloth or inference?

unsloth, with 6,084 test files; inference has 676.

The proof: [unslothai/unsloth's code at commit 98836c2](https://github.com/unslothai/unsloth/tree/98836c27662fcce34c60a6e6ca264d5973cb0058)

### Which was updated more recently, unsloth or inference?

inference: its last push was on November 14, 2026, and unsloth's on November 12, 2026.

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

### Which has more commits signed by AI, unsloth or inference?

unsloth: an AI signed 1% of its commits, with Claude Code, Gemini CLI, Cursor, Codex, and GitHub Copilot; in inference, 0.3%.

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

## Sources

- unslothai/unsloth: Read on November 12, 2026, at commit 98836c2 (https://github.com/unslothai/unsloth/tree/98836c27662fcce34c60a6e6ca264d5973cb0058)
- xorbitsai/inference: Read on November 14, 2026, at commit d2607af (https://github.com/xorbitsai/inference/tree/d2607afa6e252c8e69ee785e5be5d33216745186)

## 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.