Back to models
Intelligence Index
Max setting
26.7#36 of 83measured models
Measured · Artificial Analysis v4.3.2- Context window
- 1M tok
Performance profile
ReasoningRank #74 of 107
73.2%
GPQA DiamondCodingRank #38 of 94
48.1%
SciCodeCost efficiencyRank #61 of 67
13.2pts / $
Intelligence ÷ eval costCapabilities
Context window
Modalities
Measured setting
Reasoning trade-offs
Source (opens in a new tab)Quasar 438B
Cost per task (USD)
Quasar 438BMax, Based on GLM-5.2
- Intelligence Index
- 26.7
- Cost / task
- $2.02
- Evaluation time
- 6.2m
Pricing & access
No API offering listed in our tracked sources.
Benchmarks
Full leaderboardSource (opens in a new tab)18.7%, rank 69 of 12648.1%, rank 38 of 941.0%, rank 39 of 8776.3%, rank 47 of 1251,182 Elo, rank 38 of 101
Humanity’s Last Exam
Frontier reasoning- 1Claude Opus 5.5Max61.4%
- 2Claude Fable 5.1Max59.1%
- 3Gemini 4 ArgonHigh57.1%
- 66gpt-oss-120bHigh19.6%
- 67Gemma 4 26B A4BReasoning19.3%
- 68Gemini 3.5 Flash-LiteHigh18.8%
- 69Quasar 438BMax18.7%
- 70K-EXAONE 2.0 0803Reasoning18.6%
- 71K2 Horizon 7BReasoning18.2%
- 72G9v3-39A5BReasoning17.5%
Peers use their best measured result on this benchmark.
Alternatives
- Scores higherMiniMax-M3Reasoning
- Intelligence Index
- 29.2+2.5
- Cost / task
- $0.508−75%
- Higher score, lower costGPT-6.1 SolMax
- Intelligence Index
- 51.8+25.1
- Cost / task
- $0.724−64%
- Similar scoreNemotron 3.5 Super VLReasoning
- Intelligence Index
- 26.5−0.2
- Cost / task
- $0.188−91%
Each model at its highest-scoring measured setting · Intelligence Index v4.3.2 and its evaluation cost per task.