tencent/Hy3
SWE_Bench_Pro · 57.9 source-defined score; unit unspecified
Source-qualified evaluations, report statistics and safety observations.
SWE_Bench_Pro · 57.9 source-defined score; unit unspecified
th · 67 source-defined score; unit unspecified
mmlu_pro · 36.6 source-defined score; unit unspecified
arxiv_math · 89.6 source-defined score; unit unspecified
apex-agents · 16.6 source-defined score; unit unspecified
SWE_Bench_Pro · 47.6 source-defined score; unit unspecified
medium · 408822 source-defined score; unit unspecified
terminalbenchscience_0_1 · 9.5 source-defined score; unit unspecified
unreal_engine_windows · 71.4 source-defined score; unit unspecified
diamond · 88.27 source-defined score; unit unspecified
layout · 71.59 source-defined score; unit unspecified
prompt_level_loose · 75.9 source-defined score; unit unspecified
overall · 63.9 source-defined score; unit unspecified
ale-cli-pass-rate · 44.2 source-defined score; unit unspecified
table · 82.68 source-defined score; unit unspecified
text_formatting · 0.97 source-defined score; unit unspecified
prompt_level_loose · 76.5 source-defined score; unit unspecified
overall · 42.6 source-defined score; unit unspecified
These charts contain model-linked results. The charts above also include unlinked source profiles. Compare scores within each evaluation. Model quality, usage, transparency and system performance measure different things.