List your company

Question Answering on GPQA Diamond (Quality, Cost, and Error Reduction)

100Quality (%)Okshot
40.09655.64871.286.752Jun 25, 2026Jun 26, 2026Jun 27, 2026Jun 28, 2026

Evaluation results

MethodLinks
Okshotk=12026.061000.036100
SOTA LLM2026.06940.005—
Qwen3-235BCategory=Heuristic2026.0678.81.452—
ENTROROUTER2026.0673.20.902—
FORCCategory=Existing Routing Architectures2026.0671.70.741—
RandomCategory=Heuristic2026.0671.20.696—
Qwen3-30BCategory=Heuristic2026.0669.70.707—
RouteLLMCategory=Existing Routing Architectures2026.0665.20.271—
Qwen3-4BCategory=Heuristic2026.0662.60.052—
Router-R1Category=Existing Routing Architectures2026.0649.50.001—
xRouterCategory=Existing Routing Architectures2026.0642.40.005—