‹ PublicAI Index
The LLM benchmark aggregator.
Step 5
StepFun · 604B · open weights
Strongest in Biology research (#20 of 31), weakest in Core abilities (#153 of 200). Above par in 7 of 21 scopes. Among the models it meets almost everywhere, it finishes behind Claude Opus 5.5 and DeepSeek V4.1 Flash and ahead of Claude Sonnet 4 and GPT-4.1.
§ 1 · Profile
What it is good at
Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.
Coding52.1−14#70/1792/6
Agentic coding52.2−14.5#66/1732/5
Human preference61.3−7.5#70/3541/1
Human preference61.3−7.5#70/3541/1
Professional49−13.4#111/1942/2
Biology research46.8−23.9#20/311/1
Document design46.1−22.3#25/321/1
IT operations39.3−34.5#32/331/1
Legal53.2−12.7#65/1731/1
Finance51.9−10.5#85/1621/1
Reasoning46.7−18.7#125/1952/5
Reasoning47.5−21.5#79/1471/3
Mathematics45.4−18.4#107/1521/2
Agents49.8−16.5#129/2983/6
Long-horizon work39.8−34#24/321/1
Web research47.3−16.3#26/361/2
Knowledge work63.5−7.2#31/2082/2
Computer use37.7−30.4#61/691/2
Core abilities44.4−21.2#153/2002/4
Data analysis36.8−29.3#27/321/1
General intelligence47−22#113/1802/3
§ 2 · Head to head
What it beats, and what beats it
The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for Step 5, left for the other.
§ 3 · Sources
Where the numbers come from
6 publications, 20 figures. Every one links to the page it was read from.
LMArena Text 1448
Artificial Analysis Intelligence Index 29
GDPval-AA 1592
AA-Briefcase 1423
LiveBench 54.9LiveBench · Agentic 19.4LiveBench · Reasoning 83.8LiveBench · Coding 67.6LiveBench · Browser Use 24.8LiveBench · Data Analysis 63.5LiveBench · Research 75.5LiveBench · Design 70.8
Vals · Legal Research Bench 40.38%Vals · Harvey Legal Agent Benchmark 12.92%Vals · Finance Agent 50.67%Vals · BioMysteryBench 67.41%Vals · SRE Bench 0.38%Vals · Vibe Code Bench 70.4% (OpenHands)Vals · Code Migration 38.98%Vals · ProofBench 42%
Badge
[](https://publicai.io/model-index/m/step-5)