TarqaAI is a unified AI infrastructure and governance platform that provides a single API endpoint for over 60 AI models, including frontier reasoning models (Claude Opus 4.6, GPT-4.1, o3, Gemini 2.5 Pro), open-weight models, and self-hosted local GPUs via secure WebSocket tunnels. Key features include:
- Unified API: One endpoint for all models; switch models by changing a single string.
- Embedded RAG: Upload documents, codebases, or URLs; Tarqa handles vector search and context injection.
- GPU Tunnels: Expose local GPUs as first-class endpoints via a CLI agent.
- Governance: Dual-gate allowance (request and token limits), 5-role RBAC, SAML 2.0 SSO, and exportable audit trails.
- Observability: Per-request latency, token counts, cost breakdowns, and error traces.
- BYOK: Bring your own provider keys with $0 token markup.
- Pricing: Free tier (100 requests/mo), Starter ($15/mo), Pro ($49/mo), Team ($149/mo), and Enterprise (custom).
Target users are developers and teams building AI applications who need to manage multiple models, control costs, and maintain governance without vendor lock-in.
