MMODELYST
Capabilities/Reasoning
CAP

Reasoning

League table — best at reasoning · 30
mean percentile across this capability's benchmarks
#ModelLeagueCoverage$/1Mtok/s
1GPT-5.6 Sol (max)99.82/5$11.2585
2Claude Fable 599.82/5$2065
3Gemini 3.1 Pro Preview99.52/5$4.5140
4Claude Opus 4.899.52/5$1059
5GPT-5.6 Sol99.32/5$11.2571
6GPT-5.599.22/5$11.2572
7Muse Spark 1.199.02/5$2129
8GPT-5.6 Terra (max)98.92/5$5.63166
9GPT-5.498.72/5$5.63161
10Grok 4.598.72/5$3110
11Gemini 3.5 Flash98.62/5$3.38238
12GPT-5.6 Terra98.02/5$5.63130
13GPT-5.3 Codex97.92/5$4.8199
14GLM-5.2 (max)97.92/5$2.15200
15Claude Opus 4.797.82/5$1052
16Qwen3.7 Max97.72/5$3.75197
17Claude Sonnet 597.62/5$479
18Muse Spark97.42/5
19MiniMax-M397.42/5$0.525111
20Gemini 3 Pro Preview97.32/5$4.5
21GPT-5.6 Luna (max)97.22/5$2.25234
22Kimi K2.696.72/5$1.7143
23Claude Opus 4.696.72/5$1048
24Grok Build 0.1 061696.52/5$1.2559
25o3-pro96.31/5$3526
26GPT-5.296.22/5$4.8181
27GPT-5.6 Luna96.12/5$2.25222
28DeepSeek V4 Pro96.12/5$0.54464
29Grok 4.396.12/5$1.56152
30Gemini 3 Flash Preview95.92/5$1.13202

League = a model's mean percentile across the 5 benchmarksin this capability (best score per benchmark; partial coverage shown). Price & speed via Artificial Analysis.

Benchmarks in this capability (5)
ARC-AGI0 scores
BIG-Bench Hard1480 scores
GPQA Diamond1909 scores
MuSR1480 scores