Note: Overall leaderboard rankings may not reflect true model quality — individual benchmarks give a clearer picture. ARC-Challenge MMLU GPQA GSM8K Artificial Analysis Intelligence Index v4.1
← Back to leaderboard

StepFun: step-5-preview

3 benchmarks
CompanyStepFun 上海阶跃星辰智能科技有限公司
Country of originChina

Leaderboard position

#43
Overall rank (of 1121 models)
2638
Elo rating
3
Benchmarks
Top 3.8%
Percentile
#1
In StepFun (of 8 models)
Above: Anthropic: claude-opus-4.6 #42 Below: Thinking Machines Lab: inkling-2-small #44

Benchmark highlights

Best benchmark
Artificial Analysis Intelligence Index v4.3.2 44
Worst benchmark
Artificial Analysis Intelligence Index v5.0.1 29

Nearest competitor from another company

The model from another company with the most similar benchmark scores.

appeared as nearest in 2 of 3 benchmarks

MiniMax: minimax-m3.1-flash
2/3 66.7%
Model Votes % of benchmarks
MiniMax: minimax-m3.1-flash 2 66.7%
SpaceXAI: grok-4.6 1 33.3%

Benchmarks

Artificial Analysis Intelligence Index v4.3.2 44 Artifical Analysis Intelligence Index v5.0 29 Artificial Analysis Intelligence Index v5.0.1 29

Predicted scores

Estimated from 25 similar models and the global Elo rating. These are projections, not measured results.

Benchmark Predicted Confidence Based on
GPQA Diamond ~96 60% Meta: muse-spark-open-816b-a42b, OpenAI: gpt-6-astra
Humanity's Last Exam (no tools) ~43.37 60% DeepSeek: deepseek-v4.1-flash-1017, Anthropic: claude-haiku-5.5, OpenAI: gpt-6.1-sol
Terminal-Bench v2.1 ~83.33 60% Google: gemini-3.8-flash, Qwen: qwen3.8-max, Meta: muse-spark-open-816b-a42b
SWE-bench Pro ~70.32 60% reflection: beam-1.0, Anthropic: claude-opus-5.5
SciCode ~54.78 60% Meta: muse-spark-open-816b-a42b, OpenAI: gpt-6.1-sol, Anthropic: claude-sonnet-5.5
AA-LCR ~77.94 60% Mistral AI: Mistral Large 4 (Le Chonk), OpenAI: gpt-6-astra
Humanity's Last Exam (with tools) ~58.24 60% Meta: muse-spark-open-816b-a42b, Z.ai: glm-5.3, Anthropic: claude-haiku-5.5
MMMU-Pro ~86.44 60% Mistral AI: Mistral Large 4 (Le Chonk), OpenAI: gpt-6.1-sol, OpenAI: gpt-6-astra
BrowseComp ~85.86 60% OpenAI: gpt-6-astra
ParseBench ~76.63 60% Anthropic: claude-opus-5.5
Artificial Analysis Intelligence Index v4.1.1 (Maximum Reasoning) ~53.82 60% Google: gemini-3.8-flash, Qwen: qwen3.8-max, Anthropic: claude-fable-5.1
Artificial Analysis Intelligence Index v4.1 (Maximum Reasoning) ~45.27 60% Thinking Machines Lab: inkling
AA-Omniscience Accuracy ~36.69 54% Mistral AI: Mistral Large 4 (Le Chonk), Meta: muse-spark-open-816b-a42b, Anthropic: claude-sonnet-5.5
MCP-Atlas ~84.93 52% Meta: muse-spark-open-816b-a42b
Artificial Analysis Intelligence Index v4.3 ~40.4 52% Google: gemini-3.8-flash, Z.ai: glm-5.3, Anthropic: claude-fable-5.1
MATH-500 ~99.2 50% —
AIME 2025 ~100 50% —
AIME 2026 ~99.2 50% —
HMMT Feb 2026 ~96.9 50% —
MMLU-Pro ~90.53 50% —
IFEval ~96.75 50% —
IFBench ~82.86 50% —
HumanEval ~95.9 50% —
MBPP ~93.7 50% —
MMLU ~92.5 50% —