iamshimanto/llmswitch repository overview

⁠⚡ LLMSwitch — One Endpoint, Every AI

Open-source AI gateway (MIT). Point any OpenAI-compatible client at one local endpoint and route across 256+ providers with auto-fallback, quota-aware routing, cost controls and a full dashboard.

⁠🐳 Docker (fastest start)

docker run -d --name llmswitch -p 6000:6000 iamshimanto/llmswitch

Also on GHCR: ghcr.io/iamshimanto/llmswitch

Open http://localhost:6000⁠ — the dashboard prints a bootstrap admin key in the container logs on first run.

⁠docker compose

git clone https://github.com/iamShimanto/llmswitch.git
cd llmswitch
docker compose up -d

⁠Highlights

• OpenAI-compatible API: /v1/chat/completions, /v1/responses, /v1/messages (Anthropic native), embeddings, images, audio, videos, rerank, files, batches — SSE streaming
• 256+ providers: API-key, OAuth (Claude, Codex, Copilot, Antigravity…), keyless free tiers, cookie web-session packs (opt-in)
• 20 routing strategies: cost/latency/quota-weighted, fallback chains, session affinity, cooldowns
• Compression & caching: Caveman + RTK prompt compression, prompt cache + semantic cache
• Cost management: per-key budgets with warn/block, forecasts, CSV export
• Dashboard: analytics, playground, request inspector, PII redaction, MCP server, CLI included

Source & docs: https://github.com/iamShimanto/llmswitch⁠ — MIT License.