OpenRouter launched Fusion, an API that sends your prompt to multiple AI models in parallel, then uses a judge and synthesizer to merge the best answer. The pitch: frontier-level quality without the frontier price tag.
The timing was convenient. Anthropic had just suspended Claude Fable 5 and Mythos 5 for foreign nationals due to a US export control directive. OpenRouter stepped in the next day promising “Fable-level intelligence at half the price.”
On Perplexity’s DRACO benchmark, a budget panel — Gemini 3 Flash plus Kimi K2.6 and DeepSeek V4 Pro, synthesized by Claude Opus 4.8 — scored 64.7%. That beat solo GPT-5.5 (60%) and solo Opus 4.8 (58.8%) outright, landing within a point of Fable 5 at roughly half the cost.
It’s not a full Fable replacement. DRACO doesn’t test long-horizon work where Fable reportedly still leads. And for coding, Fusion works as a tool a coding model calls selectively, not a wholesale swap.
But the basic point holds: a panel of cheaper models, fused together, can sit right next to the most expensive solo model on results while delivering a much smaller bill. For deep research and complex planning, that’s worth knowing.
