Compare Models

Step 5 Preview

Release Date: Sep 28, 2026

Developer StepFun  🇨🇳
Context Window 1M
Max Output Tokens 1M
Token Costs (in/out) $1.00/2.70
Weights Open
Input Modalities

Accuracy

49.35 % ± 1.22

Cost / Test (Vals Index)

$ 2.626

Latency

70 min 18 s

Vals Index
BenchmarksAccuracyRankings
Vals Index

49.35%

±1.22
29/43
Code Migration

38.98%

±4.53
28/73
EMB

57.12%

±3.17
37/70
Finance Agent (v2)

50.67%

±0.32
36/74
Legal Research Bench

40.38%

±3.41
25/73
MysteryMechanism

18.47%

±2.61
21/23
ProofBench v1.1

42.00%

±4.96
31/46
Public Benefits Bench v1.1

58.12%

±1.28
34/47
Tax Agent Bench

59.54%

±3.09
34/65
Vibe Code Bench v1.1

70.40%

±4.27
34/108
BioMysteryBench

67.41%

±0.37
15/23
Harvey's Legal Agent Benchmark

9.17%

±2.44
17/74
IOI

38.06%

±6.69
34/40
Terminal-Bench 4.0

31.82%

±2.62
13/43
Terminal-Bench Science

2.86%

±2.01
29/38
Proprietary BenchmarksAcademic BenchmarksIndustry Partners
Vals
Default Provider : Stepfun
Temperature: 1
Top P: 0.95
Top K: Default
Max Output Tokens: 1,024,000