Source: https://seerepo.com/compare/OpenBMB/MiniCPM-V/vs/vllm-project/vllm-omni
Markdown source: https://seerepo.com/compare/OpenBMB/MiniCPM-V/vs/vllm-project/vllm-omni.md
Description: MiniCPM-V and vllm-omni share 25 technologies: AIOHTTP, FastAPI, cycler, fonttools, contourpy, matplotlib, Rich, bitsandbytes, einops, fairscale, onnxruntime, sentencepiece, tokenizers, torchaudio, torch, transformers, torchvision, opencv-python, accelerate, timm, pillow, decord, imageio, Pydantic, and psutil. Both are written in Python.

# MiniCPM-V vs vllm-omni

MiniCPM-V and vllm-omni share 25 technologies: AIOHTTP, FastAPI, cycler, fonttools, contourpy, matplotlib, Rich, bitsandbytes, einops, fairscale, onnxruntime, sentencepiece, tokenizers, torchaudio, torch, transformers, torchvision, opencv-python, accelerate, timm, pillow, decord, imageio, Pydantic, and psutil. Both are written in Python.

- [OpenBMB/MiniCPM-V](https://seerepo.com/r/OpenBMB/MiniCPM-V): See, hear, and interact on your local devices.
- [vllm-project/vllm-omni](https://seerepo.com/r/vllm-project/vllm-omni): A framework for efficient model inference with omni-modality models

## What they share and what sets them apart

- In common: AIOHTTP, FastAPI, cycler, fonttools, contourpy, matplotlib, Rich, bitsandbytes, einops, fairscale, onnxruntime, sentencepiece, tokenizers, torchaudio, torch, transformers, torchvision, opencv-python, accelerate, timm, pillow, decord, imageio, Pydantic, psutil
- Only in MiniCPM-V: webdataset, uvicorn, sphinx-design, sty, Gradio, seaborn, flash-attn, gradio-client, librosa, modelindex, OpenAI, pycocotools, sacrebleu, scikit-learn, sentence-transformers, umap-learn, vector-quantize-pytorch, vocos, nltk, deepspeed, peft, pycocoevalcap, spaCy, tiktoken, docutils, eva-decord, jsonlines, markdown2, myst-parser, openpyxl, sphinxcontrib-jquery, xlsxwriter, moviepy, validators
- Only in vllm-omni: lmdb, pyarrow, tensorboard-data-server, pyzmq, Hugging Face, werkzeug, auto-round-lib, facexlib, filterpy, fvcore, gguf, llvmlite, lmcache, NVIDIA, nvidia-cudnn-cu12, nvidia-cudnn-cu13, onnxruntime-rocm, open-clip-torch, openai-whisper, qwen-vl-utils, scikit-image, torchsde, trl, diffusers, safetensors, cache-dit, triton, x-transformers, s3tokenizer, kernels, torchada, markdown, markdown-it-py, markupsafe, mdurl, tifffile, av, jinja2, msgspec, opentelemetry-exporter-http-transport, opentelemetry-exporter-otlp-common, opentelemetry-exporter-prometheus, opentelemetry-semantic-conventions, opentelemetry-semantic-conventions-ai, OpenTelemetry, opentelemetry-exporter-otlp-proto-grpc

## Side by side

| | MiniCPM-V | vllm-omni |
|---|---|---|
| Size | 36,999 lines written, 105 code files | 1,649,609 lines written, 2,220 code files |
| Who writes it | 1.1% of its commits signed by AI. With Cursor and Codex | 11.7% of its commits signed by AI. With Claude Code, Cursor, GitHub Copilot, Codex, Gemini CLI, and Jules |
| Activity | Last push November 9, 2026, 646 commits, 27,172 stars | Last push November 18, 2026, 4,411 commits, 7,828 stars |
| Tests | 0 test files | 2,401 test files. 70.5% 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, MiniCPM-V or vllm-omni?

vllm-omni, with 1,649,609 lines written; MiniCPM-V has 36,999.

The proof: [vllm-project/vllm-omni's code at commit 07dd3d0](https://github.com/vllm-project/vllm-omni/tree/07dd3d0cdea3325d8026a93485b60dfe3be6f29d)

### Which has more GitHub stars, MiniCPM-V or vllm-omni?

MiniCPM-V, with 27,172 stars; vllm-omni has 7,828.

The proof: [OpenBMB/MiniCPM-V on GitHub](https://github.com/OpenBMB/MiniCPM-V)

### Which has more tests, MiniCPM-V or vllm-omni?

vllm-omni, with 2,401 test files.

The proof: [vllm-project/vllm-omni's code at commit 07dd3d0](https://github.com/vllm-project/vllm-omni/tree/07dd3d0cdea3325d8026a93485b60dfe3be6f29d)

### Which was updated more recently, MiniCPM-V or vllm-omni?

vllm-omni: its last push was on November 18, 2026, and MiniCPM-V's on November 9, 2026.

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

### Which has more commits signed by AI, MiniCPM-V or vllm-omni?

vllm-omni: an AI signed 11.7% of its commits, with Claude Code, Cursor, GitHub Copilot, Codex, Gemini CLI, and Jules; in MiniCPM-V, 1.1%.

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

## Sources

- OpenBMB/MiniCPM-V: Read on November 20, 2026, at commit 4dbbb6c (https://github.com/OpenBMB/MiniCPM-V/tree/4dbbb6c58b6f8a004d6689028417296719f0831a)
- vllm-project/vllm-omni: Read on November 18, 2026, at commit 07dd3d0 (https://github.com/vllm-project/vllm-omni/tree/07dd3d0cdea3325d8026a93485b60dfe3be6f29d)

## 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.
