Step 5 Preview
StepFun·Released Sep 18, 2026
Updated Oct 8, 11:55 PM ET
The score is on the board's scale (reference average 50, about 15 points per standard deviation), not a percentage.
By domain
Each domain is scored on the same scale as the overall score. Coverage is the share of the domain's kinds of evaluation with a result; it says how much evidence there is, not how good the model is.
Every result behind the score
The result as the evaluator publishes it, and the reading it gives on the board's scale. A hard evaluation can give a high reading for a modest result; results near 95% or 5% only say "at least" or "at most".
Coding
5 results · 64.1- SciCodeArtificial Analysis58.9%Reading 74.0
- Terminal-Bench 4.0 (AA run)Artificial Analysis33.3%Reading 70.2
- Terminal-Bench 4.0Vals AI31.8%Reading 66.3
- Code MigrationVals AI39.0%Reading 59.5
- Vibe Code BenchVals AI70.4%Reading 57.5
Research and reasoning
4 results · 54.5- Humanity's Last Exam (AA run)Artificial Analysis46.5%Reading 65.3
- ProofBenchVals AI42.0%Reading 52.5
- MysteryMechanismVals AI18.5%Reading 49.5
- Terminal-Bench ScienceVals AINear the floor2.9%Reading at most 49.0
Professional work
5 results · 57.7- Legal Research BenchVals AI40.4%Reading 62.4
- Tax Agent BenchVals AI27.3%Reading 59.1
- Harvey Legal Agent BenchmarkVals AI9.2%Reading 58.1
- Finance AgentVals AI50.7%Reading 55.0
- EMBVals AI57.1%Reading 53.6
Knowledge and accuracy
1 result · 63.6- AA-LCRArtificial Analysis88.3%Reading 66.4
Human preference
1 result · 47.1- Arena TextLMArena1,453.7Reading 41.6
Shown, not scored
Composite indices, saturated or older evaluations, evaluations still being checked, and vision, writing and multilingual results. They do not move the score.
- IOIVals AI · CodingReference only38.1%
- BioMysteryBenchVals AI · Research and reasoningWatching67.4%
- AA Intelligence IndexArtificial Analysis · Composite indicesReference only43.7
- Vals IndexVals AI · Composite indicesReference only49.4%
- Arena VisionLMArena · VisionReference only1,259.7
- Arena WebDevLMArena · Writing and designReference only1,563.9
No result yet
Scored evaluations this model has not taken. A missing result neither adds nor subtracts.
- LiveBench Coding · LiveBench
- ProgramBench · Vals AI
- Vibe Code Bench 1–100 · Vals AI
- CyberBench Patch · Vals AI
- APEX-SWE · Mercor
- FrontierMath Tiers 1–3 · Epoch AI
- FrontierMath Tier 4 · Epoch AI
- LiveBench Reasoning · LiveBench
- Chess Puzzles · Epoch AI
- Mystery Game Puzzles · Epoch AI
- APEX-Agents · Mercor
- MedCode · Vals AI
- LiveBench Data Analysis · LiveBench
- τ²-Bench Telecom (AA run) · Artificial Analysis
- τ-Bench Banking (AA run) · Artificial Analysis
- SimpleQA Verified · Epoch AI
- LiveBench Language · LiveBench
- LiveBench Instruction Following · LiveBench
- BullshitBench · BullshitBench
- IFBench (AA run) · Artificial Analysis