# kubeai-project/kubeai

AI Inference Operator for Kubernetes. The easiest way to serve ML models in production. Supports VLMs, LLMs, embeddings, and speech-to-text.

Repository: https://github.com/kubeai-project/kubeai
Canonical: https://ross.abutalabs.com/products/kubeai
Homepage: https://www.kubeai.org
Language: Go
License: Apache-2.0
License Family: permissive
Topics: k8s, llm, openai-api, autoscaler, ollama, vllm, ollama-operator, vllm-operator, ai, kubernetes, whisper, inference-operator, faster-whisper
Last push: 2026-09-03T01:03:30+00:00

## Health v2 (maintenance only)
Score: 89/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 100, release rhythm 83, longevity 74
- inputs: {"age_days": 1048, "days_push": 0, "days_rel": 34, "gap_med": 0.0, "n_releases_24m": 83}
- flags: no_readme
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1256, forks 135 (observed 2026-09-03T02:15:14.154227+00:00)

## What it is
KubeAI is a Kubernetes operator for serving machine learning models in production, supporting LLMs via vLLM and Ollama, vector embeddings, reranking, and speech-to-text with FasterWhisper. It exposes an OpenAI-compatible API and provides scale-from-zero autoscaling, prefix-aware load balancing, model caching, and LoRA adapter orchestration without requiring Istio, Knative, or other dependencies.

## Use cases
- serve llm models on kubernetes
- self-host openai-compatible inference api
- autoscale vllm replicas from zero
- run whisper speech-to-text in a cluster
- deploy embedding models on k8s
- manage lora adapters across model replicas
- load balance llm traffic with kv-cache awareness

## When to choose
- you run ML model inference on Kubernetes and want a zero-dependency operator
- you need OpenAI-compatible endpoints backed by self-hosted models
- you want scale-from-zero autoscaling without Istio or Knative
- you serve multiple model types (LLMs, embeddings, speech) from one system
- you need better load balancing than kube-proxy for stateful vLLM replicas

## When to avoid
- you don't use Kubernetes - use vLLM or Ollama directly
- you need fully managed inference - use a cloud LLM API instead
- you only run a single model on one machine
- you need training or fine-tuning - KubeAI is inference-only

## Facets
- artifact type: service
- maturity: active
- function: llm-inference, speech-recognition, machine-learning, deployment, container-orchestration, api-gateway, rag
- domain: machine-learning, large-language-models, self-hosted, cloud-computing, speech-processing
- platform: cloud, self-hosted, go
- tags: kubernetes-operator, vllm, ollama, openai-compatible-api, autoscaling, scale-from-zero, model-caching, lora-adapters, embeddings, reranking, faster-whisper, gpu-scheduling, helm-chart, prefix-aware-load-balancing, autoscaling-not-in-vocab-use-tags, kubernetes, devops, docker

## Member repositories
- kubeai-project/kubeai (main) score 89

## Provenance
- Observed fields: from GitHub, fetched 2026-09-03T02:15:14.154227+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 2026-08-30T05:07:33.310978+00:00, confidence not recorded.
  - homepage: https://www.kubeai.org (fetched 2026-08-29T12:18:19.279919+00:00, sha 29aa1f83c5b2)
  - site_page: https://www.kubeai.org/installation/aks (fetched 2026-08-29T12:18:19.289272+00:00, sha de22f5cddece)
  - site_page: https://www.kubeai.org/installation/any (fetched 2026-08-29T12:18:19.291541+00:00, sha fed6e15ec6ac)
  - site_page: https://www.kubeai.org/installation/eks (fetched 2026-08-29T12:18:19.293126+00:00, sha 8b54d1711efc)
  - site_page: https://www.kubeai.org/installation/gke (fetched 2026-08-29T12:18:19.295023+00:00, sha c4c769b76084)
  - site_page: https://www.kubeai.org/how-to/install-models (fetched 2026-08-29T12:18:19.296727+00:00, sha 941c8f7626e9)
  - site_page: https://www.kubeai.org/contributing/documentation (fetched 2026-08-29T12:18:19.298530+00:00, sha 72310b6a0d83)
- Data as of 2026-08-30T08:39:29.467469+00:00.