The AI arena is free today

Open Superagent
AI OrganizationMaintains 8 benchmarks

FutureHouse: AI benchmarks & research

Non-profit lab building AI scientists. Maintains LAB-Bench and BixBench.

Maintained by FutureHouse

8 benchmarks

In one line

FutureHouse maintains BixBench (counts toward the LLM Stats Score, 1 model scored), BixBench MCQ (counts toward the LLM Stats Score, 1 model scored), CloningScenarios (counts toward the LLM Stats Score, 1 model scored), FigQA (counts toward the LLM Stats Score, 3 models scored), LABBench2 (counts toward the LLM Stats Score, 3 models scored), LAB-Bench Practical (counts toward the LLM Stats Score, 1 model scored), ProtocolQA (counts toward the LLM Stats Score, 1 model scored) and ProtocolQA Open-Ended (counts toward the LLM Stats Score, 2 models scored). It ships no models of its own on LLM Stats.

FAQ

Common questions about FutureHouse.

What is FutureHouse?

Non-profit lab building AI scientists. Maintains LAB-Bench and BixBench. On LLM Stats it maintains 8 benchmarks, led by FigQA, LABBench2 and ProtocolQA Open-Ended. It ships no models of its own; each score on its benchmarks carries its own provenance label on the model page.

Which FutureHouse benchmarks count toward the LLM Stats Score?

FigQA, LABBench2, ProtocolQA Open-Ended, BixBench, BixBench MCQ, CloningScenarios and 2 more count toward the LLM Stats Score. Only benchmarks admitted to the core set move the main leaderboard; the rest are shown for transparency.

Which models lead FutureHouse's benchmarks?

Claude Mythos Preview leads FigQA at 89.0% (3 models scored), Gemini 4 Argon leads LABBench2 at 88.8% (3 models scored) and Grok 4.7 leads ProtocolQA Open-Ended at 70.4% (2 models scored). Each benchmark page lists the full ranking and how the results were produced.

Where are FutureHouse's official benchmark results?

FutureHouse's official site is https://www.futurehouse.org. Each benchmark page on LLM Stats links to that benchmark's own paper or leaderboard, and shows the scores next to every other benchmark so models can be compared on one page.