Fusion API: many models, one smarter answer.
Test, compare and optimize your multi-model Fusion API calls. A live playground, a cost calculator, and benchmarks for OpenRouter's Fusion — built for developers.
Free to try · runs in your browser · no signup
A multi-model panel researches, cross-reviews, and synthesizes one answer — live and free to try.
Everything to work with the Fusion API
Test it, price it, and decide when it's worth it.
Fusion Playground
Run any prompt through the Fusion API and watch the panel → judge → synthesis stream live.
Open PlaygroundCost Calculator
Estimate what a Fusion call costs vs a single model — and whether it's worth it.
Calculate costBenchmark Lab
Batch-test your own questions across Fusion and single models — compare cost, speed, and answers.
Open Benchmark LabFusion vs Fable 5
See how a multi-model panel stacks up against a single frontier model.
CompareHow Fusion works
Not one model guessing. A committee that researches, reviews, and reconciles.
Parallel research
Your prompt goes to several strong models at once. Each researches and answers independently, with the same tools.
Cross-review
A judge model reads every answer and maps the consensus, the conflicts, the gaps, and the risky claims.
Synthesis
A lead model merges the reviewed findings into one final answer — sharper than any single model alone.
One call. The whole panel.
Point any OpenAI-compatible client at the openrouter/fusion model — a panel researches in parallel, a judge cross-reviews, and one synthesized answer comes back.
Powered by a panel of frontier models
curl https://openrouter.ai/api/v1/chat/completions \
-H "Authorization: Bearer $OPENROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "openrouter/fusion",
"messages": [
{ "role": "user", "content": "Postgres or SQLite for a serverless SaaS?" }
]
}'Fusion beats frontier on deep research
Scored on DRACO — 100 deep-research tasks across 10 domains, graded on accuracy, completeness, and citation reliability.
A budget panel lands within a point of a frontier model at roughly half the cost. Most of the lift comes from the synthesis step itself — not just model diversity.
Frequently asked questions
What Fusion API is, and how it works.
Get new Fusion tools & prompt drops
New playground features, ready-to-run prompt templates, and Fusion benchmarks — straight to your inbox. No spam, unsubscribe anytime.
No spam — just new tools and prompts. Unsubscribe anytime.
Stop betting on one model.
Fusion turns a committee of models into a single, sharper answer. Try it now — free, no signup required.