BenchLeader
Menu
LeaderboardsModelsProvidersBenchmarksCompareReleases1BlogMethodology
StepFunReasoning modelNewFirst seen 19 Sept 2026 on one source; awaiting a second

Step 5 Preview

Step 5 Preview is a StepFun proprietary reasoning model. Its best configuration ranks #39 of 430 on the BenchLeader Index at 65.0 ±7.5, in the upper half. It scores highest in composite (84) and lowest in multimodal (61). At $1.43 per million tokens blended it is pricier than most ranked models.

Blended price
$1.43/M
$1.00 in · $2.70 out
Output speed
–
First answer
–
Context
1M
Released
–
date not published by our sources
How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.
  1. Overall index65
  2. Reasoning82
  3. Knowledge72
  4. Multimodal61
  5. Long context71
  6. Composite84

Benchmark results

One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

Benchmarknot statedSource
Humanity's Last Exam (AA)not in index46.5%#25Artificial Analysis
CritPt20.9%#33Artificial Analysis

Coding

Benchmarknot statedSource
SciCode (AA)not in index58.9%#8Artificial Analysis

Agents & tools

Benchmarknot statedSource
GDPval (AA)not in index53.6%#25Artificial Analysis

Knowledge

Benchmarknot statedSource
AA-Omniscience16.4#50Artificial Analysis

Multimodal

Benchmarknot statedSource
MMMU-Pro76.4%#64Artificial Analysis

Long context

Benchmarknot statedSource
AA-LCR88.3%#2Artificial Analysis

Composite

Benchmarknot statedSource
AA Intelligence Index43.6#25Artificial Analysis

Where it wins

Benchmarks where this configuration ranks in the top five of every configuration measured.

What a task costs

Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.

WorkloadTokens in / outCostWith cachingTime
Chat reply400 / 300$0.0012––
Summarise a 30-page report12,000 / 600$0.014––
Code edit6,000 / 1,500$0.010––
Agentic coding session60,000 / 4,000$0.071––
Structured extraction2,000 / 200$0.0025––

See also

Step 5 Preview vs Claude Fable 5.1Step 5 Preview vs GPT-6 AstraStep 5 Preview vs Claude Fable 5Best under $2/MCheapest 1M contextAll StepFun models

Data as of 19 Sept 2026. Compare with another model.

Cite as: BenchLeader, “Step 5 Preview: benchmarks, pricing, speed and rank”, https://www.benchleader.com/models/step-5, data as of 19 Sept 2026.