# xorbitsai/inference

Swap GPT for any LLM by changing a single line of code. Xinference lets you run open-source, speech, and multimodal models on cloud, on-prem, or your laptop — all through one unified, production-ready inference API.

Repository: https://github.com/xorbitsai/inference
Canonical: https://ross.abutalabs.com/products/inference
Homepage: https://inference.readthedocs.io
Language: Python
License: Apache-2.0
License Family: permissive
Topics: pytorch, deployment, llm, artificial-intelligence, machine-learning, whisper, inference, openai-api, gemma, llamacpp, vllm, qwen, diffusers, glm, glm-5-3, kimi, kimi-k3, llama-cpp, sglang, transformers
Last push: 2026-08-26T05:07:03+00:00

## Health v2 (maintenance only)
Score: 91/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 86, longevity 84
- inputs: {"age_days": 1176, "days_push": 7, "days_rel": 18, "gap_med": 13, "n_releases_24m": 64}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 9523, forks 863 (observed 2026-08-28T04:10:31.657160+00:00)

## What it is
Xinference is an open-source model serving platform for deploying LLMs, embedding, speech, image, and multimodal models via a unified OpenAI-compatible API. It runs on cloud, on-premises, or a laptop, with backends like vLLM, llama.cpp, SGLang, and Transformers.

## Use cases
- self-host an open-source LLM with an OpenAI-compatible API
- serve embedding models for a RAG pipeline
- run Whisper speech-to-text on my own server
- deploy a multimodal vision-language model behind one API
- swap GPT for a local model by changing one line of code
- host text-to-image generation on-premises
- run LLM inference on a laptop without cloud

## When to choose
- you need one unified API for LLMs, embeddings, audio, and image models
- you want OpenAI API compatibility with self-hosted models
- you need production-ready serving across cloud and on-prem with multiple inference backends

## When to avoid
- you only need a lightweight single-model local runner like Ollama
- you want to fine-tune or train models rather than serve them
- you need maximum single-backend performance tuning rather than flexibility

## Facets
- artifact type: service
- maturity: active
- function: llm-inference, speech-recognition, rag, api-framework, deployment, sdk
- domain: large-language-models, artificial-intelligence, machine-learning, speech-processing, self-hosted, apis
- platform: python, self-hosted, cloud, cross-platform
- tags: openai-api-compatible, model-serving, vllm, llama-cpp, sglang, multimodal, embeddings, text-to-image, whisper, on-premises, docker, gpu

## Member repositories
- xorbitsai/inference (main) score 91

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:31.657160+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:21:59.851106+00:00, confidence not recorded.
  - readme: https://github.com/xorbitsai/inference (fetched 2026-08-28T04:10:31.657160+00:00, sha e34489e49fc6)
  - homepage: https://inference.readthedocs.io (fetched 2026-08-29T08:21:45.331478+00:00, sha b74c40d7a514)
- Data as of 2026-08-30T08:39:29.467469+00:00.