Source: https://seerepo.com/compare/kyutai-labs/moshi/vs/vllm-project/vllm
Markdown source: https://seerepo.com/compare/kyutai-labs/moshi/vs/vllm-project/vllm.md
Description: moshi and vllm share 16 technologies: axum, tower, Hugging Face, Tokio, AIOHTTP, tokenizers, torch, sentencepiece, safetensors, einops, hf-hub, torchaudio, Pydantic, psutil, tracing, and tracing-subscriber. Both are written in Python.

# moshi vs vllm

moshi and vllm share 16 technologies: axum, tower, Hugging Face, Tokio, AIOHTTP, tokenizers, torch, sentencepiece, safetensors, einops, hf-hub, torchaudio, Pydantic, psutil, tracing, and tracing-subscriber. Both are written in Python.

- [kyutai-labs/moshi](https://seerepo.com/r/kyutai-labs/moshi): Moshi is a speech-text foundation model and full-duplex spoken dialogue framework. It uses Mimi, a state-of-the-art streaming neural audio codec.
- [vllm-project/vllm](https://seerepo.com/r/vllm-project/vllm): A high-throughput and memory-efficient inference and serving engine for LLMs

## What they share and what sets them apart

- In common: axum, tower, Hugging Face, Tokio, AIOHTTP, tokenizers, torch, sentencepiece, safetensors, einops, hf-hub, torchaudio, Pydantic, psutil, tracing, tracing-subscriber
- Only in moshi: axum-server, Vite, React, react-router-dom, blessed, crossterm, Ratatui, tui-logger, candle, candle-nn, mlx, candle-transformers, bitsandbytes, candle-flash-attn, julius, xformers, symphonia, ogg, env_logger, log, prometheus, tracing-appender, tracing-chrome, Tailwind CSS, webm-duration-fix, Zod
- Only in vllm: FastAPI, gitdb, pyzmq, starlette, Ray, mkdocs-material, indicatif, AMD, apache-tvm-ffi, auto-round-lib, conch-triton-kernels, cuda-bindings, cuda-python, cupy-cuda12x, cupy-cuda13x, flashinfer-python, humming-kernels, lm-format-enforcer, mooncake-transfer-engine, mooncake-transfer-engine-rocm, NVIDIA, nvidia-cudnn-frontend, OpenAI, openai-harmony, oss-harmony, quack-kernels, vllm-bench, vllm-chat, vllm-engine-core-client, vllm-llm, vllm-managed-engine, vllm-parser, vllm-proto, vllm-server, vllm-text, vllm-tokenizer, transformers, xgrammar-structural-tag, llm-multimodal, compressed-tensors, triton, mistral-common, xgrammar, partial-json-parser, torchvision, lmcache, timm, vllm-xpu-kernels, runai-model-streamer, tokenspeed-mla, llguidance, MCP, outlines-core, tpu-inference, fastokens, fastsafetensors, instanttensor, nixl, riptoken, tekken, tensorizer, tiktoken, tiktoken-rs, markdown, markupsafe, mkdocs-material-extensions, parquet, properdocs, pymdown-extensions, selectolax, xlsxwriter, pillow, image, jinja2, minijinja, torchcodec, minijinja-contrib, annotated-types, msgspec, validator, opentelemetry-semantic-conventions-ai, python-json-logger, vllm-metrics, vllm-tracing, prometheus-client, tracing-futures, model-hosting-container-standards, OpenTelemetry, prometheus-fastapi-instrumentator, tracing-log

## Side by side

| | moshi | vllm |
|---|---|---|
| Size | 37,904 lines written, 145 code files | 2,489,750 lines written, 3,450 code files |
| Who writes it | 0% of its commits signed by AI | 14% of its commits signed by AI. With Claude Code, Codex, Gemini CLI, Cursor, GitHub Copilot, Devin, Mistral Vibe, CodeRabbit, Factory, and Qwen Code |
| Activity | Last push September 9, 2026, 641 commits, 11,492 stars | Last push November 12, 2026, 25,200 commits, 96,790 stars |
| Tests | 8 test files. 2.1% of its code files have a test that imports them | 2,521 test files. 61.1% 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, moshi or vllm?

vllm, with 2,489,750 lines written; moshi has 37,904.

The proof: [vllm-project/vllm's code at commit c6573ee](https://github.com/vllm-project/vllm/tree/c6573eed5e56fd39b01d610db04616ed9b221851)

### Which has more GitHub stars, moshi or vllm?

vllm, with 96,790 stars; moshi has 11,492.

The proof: [vllm-project/vllm on GitHub](https://github.com/vllm-project/vllm)

### Which has more tests, moshi or vllm?

vllm, with 2,521 test files; moshi has 8.

The proof: [vllm-project/vllm's code at commit c6573ee](https://github.com/vllm-project/vllm/tree/c6573eed5e56fd39b01d610db04616ed9b221851)

### Which was updated more recently, moshi or vllm?

vllm: its last push was on November 12, 2026, and moshi's on September 9, 2026.

The proof: [vllm-project/vllm's commits on GitHub](https://github.com/vllm-project/vllm/commits)

### Which has more commits signed by AI, moshi or vllm?

vllm: an AI signed 14% of its commits, with Claude Code, Codex, Gemini CLI, Cursor, GitHub Copilot, Devin, Mistral Vibe, CodeRabbit, Factory, and Qwen Code.

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

## Sources

- kyutai-labs/moshi: Read on November 12, 2026, at commit e6a55d2 (https://github.com/kyutai-labs/moshi/tree/e6a55d2722a65870ef52a6c9f6ecfc0e90f38362)
- vllm-project/vllm: Read on November 12, 2026, at commit c6573ee (https://github.com/vllm-project/vllm/tree/c6573eed5e56fd39b01d610db04616ed9b221851)

## 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.