BenchLeader
Menu
Best LLMsChoose a modelCoding plansLeaderboardsModelsBenchmarksProvidersReleases7
StepFunOpen weightsReasoning modelCompare with another model

Step 5 Preview

Also searched as: Step 5, StepFun Step 5 Preview

Reasoning effort
Quality#105 of 330
54.9Index
Above the median
+4.9 on the median
Price#152 of 247
$1.43per 1M tokens
Dearer than most
1.7× the median
Speed#37 of 214
132tokens/s
Top quarter
1.7× the median

31 cheaper ranked models score higher: DeepSeek V4.1 Flash 1017 (59.2, $0.525), Qwen3.6 Plus (59.1, $1.13), DeepSeek V4.1 Flash (59.0, $0.518) and 28 more.

As of 6 Nov 2026, Step 5 Preview by StepFun scores 54.9 on the BenchLeader Index (587 ranked configurations), with its best results in long context (69) and weakest in coding (50). API pricing: $1.00 per 1M input tokens and $2.70 per 1M output tokens ($1.43 blended at 3:1), list price on 6 Nov 2026. It streams about 130 tokens/s (Artificial Analysis). Released 18 Sep 2026.

Key facts (as of 6 Nov 2026)

  • Step 5 Preview API pricing: $1.00 per 1M input tokens and $2.70 per 1M output tokens ($1.43 blended at 3:1) (list price, 6 Nov 2026).
  • Step 5 Preview streams about 130 tokens/s (Artificial Analysis, 6 Nov 2026).
  • Step 5 Preview was released on 18 Sep 2026 (StepFun).

In detail. Step 5 Preview is a StepFun open-weights reasoning model, released 18 Sep 2026. Its best configuration ranks #210 of 587 on the BenchLeader Index at 54.9 ±4.3, in the upper half. The ± is the point: 270 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in long context (69) and lowest in coding (50). At $1.43 per million tokens blended it is pricier than most ranked models. Output speed of 132 tokens per second puts it in the fastest quarter, with a first answer in 25.9 s. It has been measured at 2 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 4 Nov 2026.

Blended price
$1.43/M
$1.00 in · $2.70 out
via Artificial Analysis
Output speed
132 tok/s
measured by Artificial Analysis
First answer
26 s
first token 3.49 s
Context
1M
via Artificial Analysis
Full answer
57 s
median, reasoning included
Cost per task
$2.35
14 min a task, AA Index
Tokens per task
113k
70% of it reasoning
Released
18 Sep 2026
How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.
  1. Overall index55
  2. Reasoning53
  3. Coding50
  4. Agents & tools51
  5. Maths60
  6. Knowledge55
  7. Human preference63
  8. Multimodal59
  9. Long context69
  10. Composite41

Get an email when StepFun ships a new model

And once when it gets ranked, plus the weekly review on Mondays.

Choose providers and emails

Reasoning-effort configurations

The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.

EffortIndexRankSpeedFirst answerChat reply costCategories
high––132 tok/s26 s$0.0012Coding 60
not statedbest54.9#210132 tok/s26 s$0.0012Agents & tools 51 · Coding 50 · Composite 41 · Human preference 63 · Knowledge 55 · Long context 69 · Maths 60 · Multimodal 59 · Reasoning 53

How its index has moved

3 Oct 2026 to 5 Nov 2026
505865

The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.

Benchmark results

See all LLM benchmarks. 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.

Metadata

Benchmarknot statedSourceTrend
AA Openness Indexnot in index33.3#45Artificial Analysis

Reasoning

Benchmarknot statedSourceTrend
Arena Hard Prompts1477#71Arena (LMArena)
LiveBench 2.0 Reasoningnot in index83.8%#18LiveBench
Humanity's Last Exam (AA)not in index46.5%#44Artificial Analysis
HLE-Diamond (AA)24.3%#54Artificial Analysis
CritPtnot in index20.9%#62Artificial Analysis
CritPt v2 hard subset (AA)32.5%#62Artificial Analysis
MysteryMechanismnot in index18.5%#26Vals AI

Coding

Benchmarkhighnot statedSourceTrend
SciCodeeffort not stated–58.9%#18SciCode
Arena Codingeffort not stated–1506#67Arena (LMArena)
Arena WebDev1571#41–Arena (LMArena)
LiveBench 2.0 Codingnot in indexeffort not stated–85.6%#16LiveBench
SciCode (AA)not in indexeffort not stated–58.9%#18Artificial Analysis
IOIeffort not stated–38.1%#39Vals AI
Code Migrationnot in indexeffort not stated–39.0%#32Vals AI
Code Migration v1.1not in indexeffort not stated–47.0%#17Vals AI
SWE-bench Multimodalnot in indexeffort not stated–40.8%#17Vals AI
Vibe Code Bench v1.1not in indexeffort not stated–70.4%#47Vals AI
Terminal-Bench 4.0 (AA)not in indexeffort not stated–33.3%#47Artificial Analysis
FrontierCode 1.1 (AA)not in indexeffort not stated–38.4%#42Artificial Analysis
FrontierCode 2.0 (AA)effort not stated–12.0%#77Artificial Analysis

Agents & tools

Benchmarknot statedSourceTrend
Arena Agent-2.9#37Arena (LMArena)
LiveBench Long Agenticnot in index19.4%#23LiveBench
LiveBench Browsernot in index24.8%#25LiveBench
GDPval-AA v2.1, normalised scorenot in index54.3%#26Artificial Analysis
GDPval-AA v2.1not in index1594#28Artificial Analysis
ITBench SRE (AA)not in index55.6%#2Artificial Analysis
Analyst Agent (AA)not in index35.0%#30Artificial Analysis
APEX-Agents (AA)not in index38.0%#5Artificial Analysis
Finance Agent v2not in index50.7%#43Vals AI
Finance Agent v2.1not in index58.9%#25Vals AI
Harvey's Legal Agent Benchmark v1.1not in index12.9%#15Vals AI
Harvey's Legal Agent Benchmark v1not in index9.2%#22Vals AI
Legal Research Benchnot in index40.4%#33Vals AI
SREBenchnot in index0.4%#32Vals AI
Tax Agent Benchnot in index27.3%#37Vals AI
Terminal-Bench 4.0 (Vals)not in index31.8%#22Vals AI
Terminal-Bench Sciencenot in index2.9%#33Vals AI
AutomationBenchnot in index51.0%#92Artificial Analysis
GDP.pdf (AA)not in index14.8%#114Artificial Analysis
EnterpriseOps-Gymnot in index47.2%#10Artificial Analysis
AA-Briefcase v1.1not in index1425#55Artificial Analysis
Terminal-Bench-Science 0.1 (AA)not in index2.4%#85Artificial Analysis
CWE-Bench (AA)52.5%#14Artificial Analysis
DeepSec-Bench (AA)15.8%#15Artificial Analysis
CyberGym end-to-end (AA)74.8%#7Artificial Analysis

Maths

Benchmarknot statedSourceTrend
ProofBench42.0%#39Vals AI
Arena Maths1466#61Arena (LMArena)

Knowledge

Benchmarknot statedSourceTrend
LiveBench 2.0 Data Analysisnot in index63.5%#26LiveBench
LiveBench Researchnot in index72.8%#19LiveBench
AA-Omniscience v1.116.4#50Artificial Analysis
BioMysteryBenchnot in index67.4%#19Vals AI
Excel Modeling Benchmarknot in index57.1%#42Vals AI
Excel Modeling Benchmark v1.1not in index69.5%#17Vals AI
Public Benefits Bench v1.1not in index58.1%#37Vals AI
AA-Omniscience v1.1: accuracynot in index36.4%#72Artificial Analysis
AA-Omniscience v1.1: non-hallucinationnot in index68.7%#31Artificial Analysis

Instruction following

Benchmarknot statedSourceTrend
Arena Instruction Followingnot in index1446#72Arena (LMArena)

Human preference

Benchmarknot statedSourceTrend
Arena Text1448#81Arena (LMArena)
Arena Creative Writingnot in index1411#90Arena (LMArena)
Arena Multi-turnnot in index1442#100Arena (LMArena)
Arena Longer Queriesnot in index1460#77Arena (LMArena)

Multimodal

Benchmarknot statedSourceTrend
Arena Vision1265#49Arena (LMArena)
LiveBench Designnot in index76.9%#17LiveBench
MMMU-Pro76.4%#98Artificial Analysis
Code2Video-AAnot in index821#24Artificial Analysis
AA-3Dnot in index854#20Artificial Analysis

Long context

Benchmarknot statedSourceTrend
AA-LCR v1.188.3%#2Artificial Analysis
MLCRnot in index18.3%#62Artificial Analysis

Composite

Benchmarknot statedSourceTrend
LiveBench 2.058.6%#24LiveBench
AA Intelligence Index v5.0not in index28.5#61Artificial Analysis
AA Intelligence Index v4.3.2not in index43.7#49Artificial Analysis
AA Progress Index v1not in index318.2#70Artificial Analysis
Vals Indexnot in index54.8#22Vals AI
Vals Finance Indexnot in index55.4%#26Vals AI
Vals Science Indexnot in index29.6%#19Vals AI
AA Cyber Indexnot in index47.7#9Artificial Analysis

Where it wins

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

API pricing: where to run it

Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.

ProviderSpeedFirst tokenInput $/MOutput $/MContextQuantisation
StepFun105 tok/s3.49 s$1.00$2.701Mfp8
Featherless24 tok/s3.97 s$1.00$2.70262kfp4

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 at $0.105 per 1M. Reasoning tokens are not modelled.

WorkloadTokens in / outCostWith cachingTime
Chat reply400 / 300$0.0012$0.000928.1 s
Summarize a 30-page report12,000 / 600$0.014$0.005630.4 s
Code edit6,000 / 1,500$0.010$0.006037.2 s
Agentic coding session60,000 / 4,000$0.071$0.03056.1 s
Structured extraction2,000 / 200$0.0025$0.001227.4 s

More from StepFun

New this month

Compare Step 5 Preview with…

See also

Most usedBest under $2/MBest open weightsCheapest 1M contextAll StepFun models

Embed or cite

Show where Step 5 Preview ranks on a model card or README. The badge redraws itself with every daily update, carries the data date and links back to this page. To see ranks like this on OpenRouter and Hugging Face model pages, get the Chrome extension.

[![Step 5 Preview on BenchLeader: #104 overall](https://www.benchleader.com/badge/step-5.svg)](https://www.benchleader.com/models/step-5)

Cite this

The ranking changes daily, so a reference carries the data date. APA style:

BenchLeader. (2026, November 6). Step 5 Preview on the BenchLeader Index (method v1.4) [Data set]. Retrieved November 6, 2026, from https://www.benchleader.com/models/step-5

BibTeX:

@misc{benchleader_step_5_20261106,
  author       = {{BenchLeader}},
  title        = {{Step 5 Preview on the BenchLeader Index (method v1.4)}},
  year         = {2026},
  month        = nov,
  howpublished = {\url{https://www.benchleader.com/models/step-5}},
  note         = {Data as of 2026-11-06}
}

Every badge style, the embeddable top-10 widget and how to cite the Index

Questions people ask

What are Step 5 Preview's benchmark scores?
As of 6 Nov 2026, Step 5 Preview scores 54.9 ±4.3 on the BenchLeader Index, strongest in long context (69) and weakest in coding (50). That rests on independent benchmark results only, and every score on this page links to its publisher.
How does Step 5 Preview rank?
Step 5 Preview is #210 of 587 ranked configurations on 6 Nov 2026. 270 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 69.3.
How much does Step 5 Preview cost per 1M tokens?
As of 6 Nov 2026, Step 5 Preview costs $1.00 per 1M input tokens and $2.70 per 1M output tokens, $1.43 blended (three input tokens per output token).
Is Step 5 Preview better than Qwen3.8-Flash-Next?
As of 6 Nov 2026, Qwen3.8-Flash-Next and Step 5 Preview are neck and neck on quality (54.6 vs 54.9 on the BenchLeader Index, inside the ±10.4 margin of error), and Qwen3.8-Flash-Next costs 6.2× less, which makes it the better buy for most workloads. The head-to-head page has the benchmark-by-benchmark detail.
When was Step 5 Preview released?
StepFun released Step 5 Preview on 18 Sep 2026. BenchLeader has results for it at 2 reasoning-effort settings.

Data as of 6 Nov 2026. Compare these configurations.

Cite as: BenchLeader, “Step 5 Preview: benchmarks, pricing, speed and rank”, data as of 6 Nov 2026, https://www.benchleader.com/models/step-5.