GoModel

GoModel is an open-source, high-performance AI gateway written in Go that provides a unified OpenAI-compatible API across multiple LLM providers including OpenAI, Anthropic, Gemini, Groq, xAI, Azure OpenAI, Oracle, Ollama, and OpenRouter. It serves as a LiteLLM alternative with built-in observability, guardrails, streaming support, and a two-layer response caching system. Deployable via Docker in seconds, GoModel automatically detects available providers based on supplied credentials and exposes a consistent API surface for all of them.

• Unified OpenAI-Compatible API: Route requests to any supported provider using a single consistent API endpoint, eliminating provider-specific SDK changes.
• Multi-Provider Support: Supports OpenAI, Anthropic, Gemini, Groq, xAI, Azure OpenAI, Oracle, Ollama, OpenRouter, and Z.ai with chat completions, embeddings, files, and batch operations.
• Two-Layer Response Cache: Exact-match caching via Redis for sub-millisecond repeated request handling, plus semantic caching using vector search (Qdrant, pgvector, Pinecone, Weaviate) for ~60–70% hit rates on similar queries.
• Observability & Metrics: Built-in Prometheus metrics endpoint, audit logging, usage analytics dashboard, and per-model usage breakdowns via admin API.
• Guardrails Pipeline: Configurable guardrails for request and response filtering before output reaches the client.
• Provider-Native Passthrough: Pass requests directly to upstream providers via /p/{provider}/... routes for full native API access.
• Flexible Storage Backends: Supports SQLite, PostgreSQL, and MongoDB for usage logs, audit trails, and batch metadata.
• Easy Deployment: Deploy with a single docker run command; configure entirely via environment variables or an optional config.yaml.
• Admin Dashboard: Built-in web UI for monitoring token usage, estimated costs, model analytics, and audit conversation threads.
• Streaming Support: Full streaming support for chat completions across all compatible providers.