The open-source model router that learns from every request, runs on your infrastructure, and puts every AI model at your fingertips — better than Sakana Fugu, zero vendor lock-in, zero monthly fees.
Everything Fugu does, plus what it can't.
Three strategies — similarity, cost, and fallback — pick the best model for every prompt in under 20ms.
Learns from every request. Auto-retrains after every 10 requests. Gets smarter over time.
Every routing decision is visible. No black box. See which model was picked and why.
Run entirely on-premise with local models via Ollama. Air-gapped capable. No cloud dependency.
Real-time web dashboard showing routing decisions, model usage, costs, and latency.
Drop-in replacement for OpenAI API. Works with any OpenAI SDK. 100+ providers via LiteLLM.
See why developers are switching to open source.
| Feature | Sakana Fugu | Fugusashi |
|---|---|---|
| Model Routing | ✓ Proprietary | ✓ Open & transparent |
| Multi-Agent Orchestration | ✓ Fugu Ultra | ✓ Phase 2 |
| Self-Hosting | ❌ Cloud-only | ✓ Local-first |
| Cost | $5-30/M tokens | ✓ Free |
| Transparency | ❌ Black box | ✓ Every decision visible |
| Feedback Loop | ❌ Static | ✓ Learns from every request |
| Model Pool | ❌ Fixed by Sakana | ✓ You control |
| Training Data | ❌ Proprietary | ✓ Community + your traffic |
| License | Proprietary | ✓ MIT |
Up and running in 5 minutes.
Two tiers. One fast router. One deep orchestrator.
The learning brain. Evolves to pick the best model for every prompt.
Inspired by Sakana AI's TRINITY paper. Uses CMA-ES (Covariance Matrix Adaptation Evolution Strategy) to evolve routing weights. Runs locally in under 1ms per decision. Gets smarter from every request.
Learns from every request. Gets smarter over time.
Outcomes are recorded automatically. After every 10 requests, the similarity router retrains. Over time it learns which prompts work best with which models — without any manual intervention.
Collaborative routing without data sharing.
Each organization trains locally on their own data. Only weight updates (with differential privacy noise) are shared. The result is a router smarter than any single deployment.
Every routing decision comes with a natural language explanation.
Research contributions: federated routing learning, human-interpretable routing, CMA-ES adaptation.