List your company

General AI Assistant Tasks on GAIA Level 3 original (test)

37.5PerformanceClaude-4
-1.58.62518.7528.875Jan 6, 2026

Evaluation results

MethodLinks
Claude-4Framework=CK-Pro, Setting=Manual2026.0137.570.2212.96
EvoRouteFramework=CK-Pro, Setting=Ours2026.0133.3325.314.35
Qwen3-14bFramework=Smolagent, Setting=Manual2026.0127.390.14.4
EvoRouteFramework=Smolagent, Setting=Ours2026.0127.045.71.2
GPT-4.1Framework=CK-Pro, Setting=Manual2026.0126.779.624.71
PromptLLMFramework=CK-Pro, Setting=Routing2026.0124.3916.6415.89
MasRouterFramework=CK-Pro, Setting=Routing2026.0123.5114.9515
GraphRouterFramework=CK-Pro, Setting=Routing2026.0122.9415.514.22
Claude-4Framework=Smolagent, Setting=Manual2026.0122.9414.11.8
GPT-4.1Framework=Smolagent, Setting=Manual2026.0116.013.52.1
GPT-4oFramework=CK-Pro, Setting=Manual2026.0115.2611.446.58
PromptLLMFramework=Smolagent, Setting=Routing2026.0115.044.51.9
MasRouterFramework=Smolagent, Setting=Routing2026.0115.044.51.8
GraphRouterFramework=Smolagent, Setting=Routing2026.0115.0341.8
Qwen3-14bFramework=CK-Pro, Setting=Manual2026.0100.523.54