ModelCap
Current AI model comparison
Exact identities · public evidence
Qwen
Qwen3 Coder 30B A3B Instruct
Position
#93
Index
52.4
Estimated
publisher-corpus-prior over 203 held-out anchors (7%); launch card against 3 resolved peers on 4 rows (94%); no cross-lab optimism probe available; shrunk 0.1 up toward the measured corpus
Comparison side 1
Z.ai
GLM 5 Turbo
Position
#92
Index
52.6
Measured
publisher-corpus-prior over 203 held-out anchors (64%); 1 specialist-board observation at 1.2% support (36%); shrunk 23.8 up toward the measured corpus
Comparison side 2
Output / 1M
$0.28
vs $4.00
Context
262K
vs 203K
Weights
Open weights
vs API only