Model

Qwen3-VL-8B-Instruct

Alibaba · open weights

qwenlm.github.io

Knowledge & Reasoning

Knowledge & Reasoning
BenchmarkScoreMetricHarnessSourceConfidenceLink
WorldFact-Bench8.1accuracydirect judgmentIndependentmediumarxiv.org

Coding

Coding
BenchmarkScoreMetricHarnessSourceConfidenceLink
Hob-VL50.6accuracyN/AIndependentmediumarxiv.org

Agentic / Tool Use

Agentic / Tool Use
BenchmarkScoreMetricHarnessSourceConfidenceLink
CASE longitudinal multimodal benchmark47.1overall_(%)N/AIndependentmediumarxiv.org
PlaySuite0.7mean_progress_score_(no_history)N/AIndependentmediumarxiv.org

Multimodal

Multimodal
BenchmarkScoreMetricHarnessSourceConfidenceLink
AgroGround64.4accuracyzero-shotIndependentmediumarxiv.org
BrainTRACE33.7accuracyN/AIndependentmediumarxiv.org
CROSS-Bench68.0accuracyN/AIndependentmediumarxiv.org
DataVista46.2accuracyN/AIndependentmediumarxiv.org
EgoExo-Next36.3accuracyN/AIndependentmediumarxiv.org
EgoTools-Bench46.1egotools-bench_overallN/AIndependentmediumarxiv.org
FG-Sudoku0.4accuracyN/AIndependentmediumarxiv.org
G-RSVQA77.5accuracyN/AIndependentmediumarxiv.org
GeoSpatialBench37.2accuracyN/AIndependentmediumarxiv.org
InstanceBench52.0[email protected]N/AIndependentmediumarxiv.org
InteractionBench34.8scoreN/AIndependentmediumarxiv.org
Knossos46.1f1N/AIndependentmediumarxiv.org
MVGraph5.4accuracyN/AIndependentmediumarxiv.org
NegT2IBench74.0spearman_ρN/AIndependentmediumarxiv.org
OccludeBench27.7accuracyN/AIndependentmediumarxiv.org
OmniVCBench34.3accuracyN/AIndependentmediumarxiv.org
PhysicsLENS0.7plausibility_scoreN/AIndependentmediumarxiv.org
R-GroundBench64.9accuracyN/AIndependentmediumarxiv.org
RoadSceneVQA-G52.2f1N/AIndependentmediumarxiv.org
SpaceConflict51.0accuracyN/AIndependentmediumarxiv.org
SpatialChain78.9accuracyN/AIndependentmediumarxiv.org
SQUARE-Bench87.1overallN/AIndependentmediumarxiv.org
UkrMeme0.3hf1N/AIndependentmediumarxiv.org
Unified Suite of 25 Visual Reasoning Benchmarks61.1mean_scoreN/AIndependentmediumarxiv.org
Video-Index9.5accuracyN/AIndependentmediumarxiv.org

Safety & Alignment

Safety & Alignment
BenchmarkScoreMetricHarnessSourceConfidenceLink
OR-Bench58.0accuracyN/AIndependentmediumarxiv.org

Reproduced by the community

Loading community runs…