NEWHow AI models draw the world map · Gecko Tests
Beta
Home/Models/Step 5 Preview
stepfun logo

Step 5 Preview

by stepfun · Released Sep 2026

Multimodal1M ContextPreview
Compare
Context
1.0M tokens (~524 books)
Input $/1M
$1.00
Output $/1M
$2.70
Type
multimodal
License
Proprietary
Benchmarks
7 tested
Data as of
About

Step 5 Preview is StepFun's flagship model for agentic work, built on a sparse Mixture-of-Experts architecture (27B active / 600B total parameters). It performs strongly in software engineering and professional...

Tested on 7 benchmarks. Top scores: Artificial Analysis · Long Context Reasoning (88.3%), Artificial Analysis · MMMU Pro (76.4%), Artificial Analysis · SciCode (58.9%).

Capabilities
speed
55.6
#19 globally
Benchmark Scores
Compare All
Tested on 7 benchmarks · Ranked across 1 categories
Score Distribution (all 319 models)
0255075100
speedCompare speed →
Artificial Analysis · Long Context Reasoning
88.3·
Artificial Analysis · MMMU Pro
76.4·
Artificial Analysis · SciCode
58.9·
Quick compare:Alibaba logovs Qwen3 Coder Next FP8Alibaba logovs Qwen2.5 Coder 32B Instruct AWQvs Speaker Diarization Community 1
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Recently Happened
New model: Step 5 Preview by stepfun
Oct 15, 2026
Similar Models
Alibaba logo
Qwen3 Coder Next FP8
Alibaba
n/an/a
Alibaba logo
Qwen2.5 Coder 32B Instruct AWQ
Alibaba
n/an/a
Speaker Diarization Community 1
Pyannote
n/an/a
Links
Info
stepfunPricing explorerDevelopers · API
Research
Technical Report
Documentation
API DocsPlayground
Community
@stepfun
BenchGecko API
step-5-preview
Specifications
  • Typemultimodal
  • Context1.0M tokens (~524 books)
  • ReleasedSep 2026
  • LicenseProprietary
  • Statuspreview
  • Cost / Message~$0.005
Available On
stepfun logostepfun$1.00
Categories
speed
Learn More
context-windowtransformertokens
Share & Export
Tweet
Related Models
Alibaba logoQwen3 Coder Next FP8
n/a
Alibaba logoQwen2.5 Coder 32B Instruct AWQ
n/a
Speaker Diarization Community 1
n/a
Jais 3 0.9B
n/a
Step 5 Preview is a proprietary multimodal AI model by stepfun, released in September 2026. Context window: 1M tokens.

Related Models

Qwen3 Coder Next FP8 · AlibabaQwen2.5 Coder 32B Instruct AWQ · AlibabaSpeaker Diarization Community 1 · PyannoteJais 3 0.9B · MBZUAIGemma 3 12B (free) · Google DeepMind

Benchmarks

Artificial Analysis · Long Context ReasoningArtificial Analysis · MMMU ProArtificial Analysis · SciCodeArtificial Analysis · GDPvalArtificial Analysis · Humanity's Last Exam

Related Pages

stepfun · Providerstepfun · EconomyAll ModelsCompare ModelsPricingDevelopers · APIContext Window · Glossary

Key facts · as of 2026-10-16

  • Step 5 Preview by stepfun. Not enough public benchmark scores to rank yet.
  • List price $1.00 input · $2.70 output per 1M tokens (as of 2026-10-16).
  • Sold by 4 providers (as of 2026-10-16): DeepInfra (fp8) $1.00 in / $2.70 out · Featherless (fp4) $1.00 in / $2.70 out · InferenceNet (fp4) $1.00 in / $2.70 out · StepFun (fp8) $1.00 in / $2.70 out. Every provider

How to cite · data as of 2026-10-16

Step 5 Preview · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-16. https://benchgecko.ai/model/step-5-preview

Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP