As of October 21, 2026, StepFun Step 5 Preview ranks #41 of 123 models overall with 60.5 ±10.7 on the LLMBenchmark Index. Step 5 Preview ranks #36 of 122 in reasoning, its strongest category. StepFun Step 5 Preview costs $1.00 per million input tokens and $2.70 per million output tokens. Step 5 Preview was released on October 8, 2026.
Step 5 Preview's output price ranks 219 of 349 priced models, counting from the cheapest.
Listed on OpenRouter October 8, 2026, 13 days before our data. Results so far on 61 of 195 public boards; this page updates every day as boards publish.
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...
stepfun/step-5-preview The tick on a coding bar is the median of every model measured on it. Outside coding, the place is this model's rank on that board. A board that never ran the model says so.
Measured in 4 of 7 categories. A place counts only the models measured on that board.
The index averages only the coding benchmarks that have a result; the others are skipped, never filled in. How it is weighted
Not measured on Aider polyglot, SWE-bench Verified, SWE-bench Lite, SWE-bench (full test set), SWE-bench Multimodal, SWE-bench Multilingual, AA Coding Index, LiveCodeBench (Vals AI), ProgramBench (Vals AI), Terminal-Bench 2.1 (Vals AI), Vibe Code Bench 1-100 (Vals AI), SkillsBench (Vals AI), CyberBench (Vals AI), LiveCodeBench (Artificial Analysis run), Terminal-Bench Hard (Artificial Analysis run), Terminal-Bench, DeepSWE, FrontierSWE, GSO-Bench, WeirdML, MirrorCode, Cybench, ALE-Bench, SciCode
Not measured on Epoch Capabilities Index
Not measured on AA Agentic Index, Design Arena: agents building web apps, Design Arena: agents building full-stack apps, Design Arena: agents building mobile apps, Design Arena: agents building Android apps, Design Arena: agents building Godot games, Design Arena: agents building games, Design Arena: HTML slides, Design Arena: PowerPoint slides, Design Arena: spreadsheets, Harvey's Legal Agent Benchmark (Vals AI), Time Horizon Index (Vals AI), CUA-bench (Vals AI), Robo-bench (Vals AI), τ²-Bench Telecom, APEX-Agents, METR time horizons, Remote Labor Index, GDPval, OSWorld 2.0, BALROG, DeepResearch Bench, PostTrainBench, TheAgentCompany, OSWorld, SpireClimb, Vending-Bench 2, LMArena Search, Berkeley Function Calling Leaderboard (BFCL), BrowseComp-Plus
Not measured on GPQA Diamond, MMMU Pro (Vals AI), GPQA Diamond (Artificial Analysis run), GPQA Diamond (Epoch AI run), ARC-AGI, ARC-AGI-2, Humanity's Last Exam, SimpleBench, Chess puzzles, Mystery game puzzles, EBR-bench, VPCT, BIG-Bench Hard, Kagi LLM Benchmark
Not measured on AIME, MATH 500 (Vals AI), MGSM (Vals AI), AA Math Index, AIME 2025 (Artificial Analysis run), OTIS Mock AIME, FrontierMath, tiers 1 to 3, FrontierMath, tier 4, MATH level 5, GSM8K
Not measured on MMLU Pro, LegalBench (Vals AI), CaseLaw (Vals AI), CorpFin (Vals AI), TaxEval (Vals AI), MortgageTax (Vals AI), MedQA (Vals AI), MedCode (Vals AI), MMLU Pro (Artificial Analysis run), SimpleQA Verified, MMLU, GeoBench, AI2 Reasoning Challenge, FACTS Benchmark Suite
Not measured on Design Arena: websites, Design Arena: UI components, Design Arena: game dev, Design Arena: data viz, Design Arena: 3D, Design Arena: all code categories, Design Arena: SVG, Design Arena: ASCII art
Not measured on Vals Index, SAGE (Vals AI), MedScribe (Vals AI), Vals Multimodal Index, IFBench, DTBench, LMCA, Fiction.LiveBench, Creative writing (Lech Mazur), CL-bench, Surface Evolver Bench, Furniture assembly, ExploitBench, HellaSwag, WinoGrande, PIQA, CL-bench Life, LMArena Text: German, LMArena Text: Korean, LMArena Text: Polish, LMArena Vision: Chinese, LMArena Vision: humor, LMArena Vision: captioning, LMArena Vision: recognizing things, LMArena Vision: creative writing, LMArena Document
Not measured on Design Arena: images, Design Arena: graphic design, Design Arena: logos, LMArena Text to Image, LMArena Image Edit
Public boards shown as published (Vals AI, Artificial Analysis, Design Arena) and kept out of the coding index. Each links to its source on the benchmark page.
| Benchmark | Result | Index weight |
|---|---|---|
| Aider polyglotCoding | Not measured | Not in index |
| SWE-bench VerifiedCoding | Not measured | Not in index |
| SWE-bench LiteCoding | Not measured | Not in index |
| SWE-bench (full test set)Coding | Not measured | Not in index |
| SWE-bench MultimodalCoding | Not measured | Not in index |
| SWE-bench MultilingualCoding | Not measured | Not in index |
| LiveBench codingCoding | 85.6% | 15% |
| LMArena WebDevCoding | 1567 | 10% |
| Vals IndexGeneral | Not measured | Not in index |
| GPQA DiamondReasoning | Not measured | Not in index |
| AIMEMath | Not measured | Not in index |
| MMLU ProKnowledge | Not measured | Not in index |
| AA Intelligence IndexIntelligence | 43.7 | Not in index |
| AA Coding IndexCoding | Not measured | Not in index |
| AA Agentic IndexAgentic work | Not measured | Not in index |
| Design Arena: websitesWeb design | Not measured | Not in index |
| Design Arena: UI componentsWeb design | Not measured | Not in index |
| Design Arena: game devWeb design | Not measured | Not in index |
| Design Arena: data vizWeb design | Not measured | Not in index |
| Design Arena: 3DWeb design | Not measured | Not in index |
| Design Arena: all code categoriesWeb design | Not measured | Not in index |
| Design Arena: SVGWeb design | Not measured | Not in index |
| Design Arena: ASCII artWeb design | Not measured | Not in index |
| Design Arena: agents building web appsAgentic work | Not measured | Not in index |
| Design Arena: agents building full-stack appsAgentic work | Not measured | Not in index |
| Design Arena: agents building mobile appsAgentic work | Not measured | Not in index |
| Design Arena: agents building Android appsAgentic work | Not measured | Not in index |
| Design Arena: agents building Godot gamesAgentic work | Not measured | Not in index |
| Design Arena: agents building gamesAgentic work | Not measured | Not in index |
| Design Arena: HTML slidesAgentic work | Not measured | Not in index |
| Design Arena: PowerPoint slidesAgentic work | Not measured | Not in index |
| Design Arena: spreadsheetsAgentic work | Not measured | Not in index |
| Design Arena: imagesImage generation | Not measured | Not in index |
| Design Arena: graphic designImage generation | Not measured | Not in index |
| Design Arena: logosImage generation | Not measured | Not in index |
| LiveBench reasoningReasoning | 83.8% | Not in index |
| LiveBench data analysisGeneral | 63.5% | Not in index |
| LiveBench researchAgentic work | 72.8% | Not in index |
| LiveBench browserAgentic work | 24.0% | Not in index |
| LiveBench long agenticAgentic work | 27.5% | Not in index |
| LiveBench designWeb design | 76.9% | Not in index |
| LiveCodeBench (Vals AI)Coding | Not measured | Not in index |
| SWE-bench Verified (Vals AI)Coding | Not measured | 15% |
| Terminal-Bench 4 (Vals AI)Coding | 31.8% | 25% |
| Vibe Code Bench (Vals AI)Coding | 70.4% | 20% |
| Code migration (Vals AI)Coding | 39.0% | Not in index |
| ProgramBench (Vals AI)Coding | Not measured | Not in index |
| SREBench (Vals AI)Agentic work | 0.4% | Not in index |
| IOI (Vals AI)Coding | 38.1% | Not in index |
| Terminal-Bench 2.1 (Vals AI)Coding | Not measured | Not in index |
| SWE-bench Multimodal (Vals AI)Coding | 40.8% | Not in index |
| Vibe Code Bench 1-100 (Vals AI)Coding | Not measured | Not in index |
| SkillsBench (Vals AI)Coding | Not measured | Not in index |
| CyberBench (Vals AI)Coding | Not measured | Not in index |
| Finance Agent (Vals AI)Agentic work | 50.7% | Not in index |
| Tax Agent Bench (Vals AI)Agentic work | 27.3% | Not in index |
| Legal Research Bench (Vals AI)Agentic work | 40.4% | Not in index |
| Harvey's Legal Agent Benchmark (Vals AI)Agentic work | Not measured | Not in index |
| Excel Modeling Benchmark (Vals AI)Agentic work | 57.1% | Not in index |
| BioMysteryBench (Vals AI)Agentic work | 67.4% | Not in index |
| Terminal-Bench-Science (Vals AI)Agentic work | 2.9% | Not in index |
| Vals Science IndexAgentic work | 29.6% | Not in index |
| Time Horizon Index (Vals AI)Agentic work | Not measured | Not in index |
| CUA-bench (Vals AI)Agentic work | Not measured | Not in index |
| Robo-bench (Vals AI)Agentic work | Not measured | Not in index |
| Public Benefits Bench (Vals AI)General | 58.1% | Not in index |
| MysteryMechanism (Vals AI)Reasoning | 18.5% | Not in index |
| MMMU Pro (Vals AI)Reasoning | Not measured | Not in index |
| MATH 500 (Vals AI)Math | Not measured | Not in index |
| MGSM (Vals AI)Math | Not measured | Not in index |
| SAGE (Vals AI)General | Not measured | Not in index |
| LegalBench (Vals AI)Knowledge | Not measured | Not in index |
| CaseLaw (Vals AI)Knowledge | Not measured | Not in index |
| CorpFin (Vals AI)Knowledge | Not measured | Not in index |
| TaxEval (Vals AI)Knowledge | Not measured | Not in index |
| MortgageTax (Vals AI)Knowledge | Not measured | Not in index |
| MedQA (Vals AI)Knowledge | Not measured | Not in index |
| MedCode (Vals AI)Knowledge | Not measured | Not in index |
| MedScribe (Vals AI)General | Not measured | Not in index |
| Vals Multimodal IndexGeneral | Not measured | Not in index |
| AA Math IndexMath | Not measured | Not in index |
| GPQA Diamond (Artificial Analysis run)Reasoning | Not measured | Not in index |
| Humanity's Last Exam (Artificial Analysis run)Reasoning | 46.5% | Not in index |
| MMLU Pro (Artificial Analysis run)Knowledge | Not measured | Not in index |
| LiveCodeBench (Artificial Analysis run)Coding | Not measured | Not in index |
| SciCodeCoding | 58.9% | Not in index |
| Terminal-Bench Hard (Artificial Analysis run)Coding | Not measured | Not in index |
| IFBenchGeneral | Not measured | Not in index |
| Long context reasoning (AA-LCR)General | 88.3% | Not in index |
| τ²-Bench TelecomAgentic work | Not measured | Not in index |
| AIME 2025 (Artificial Analysis run)Math | Not measured | Not in index |
| Terminal-BenchCoding | Not measured | Not in index |
| DeepSWECoding | Not measured | Not in index |
| FrontierCodeCoding | Not measured | 15% |
| FrontierSWECoding | Not measured | Not in index |
| GSO-BenchCoding | Not measured | Not in index |
| WeirdMLCoding | Not measured | Not in index |
| MirrorCodeCoding | Not measured | Not in index |
| APEX-AgentsAgentic work | Not measured | Not in index |
| METR time horizonsAgentic work | Not measured | Not in index |
| Remote Labor IndexAgentic work | Not measured | Not in index |
| GDPvalAgentic work | Not measured | Not in index |
| OSWorld 2.0Agentic work | Not measured | Not in index |
| BALROGAgentic work | Not measured | Not in index |
| DeepResearch BenchAgentic work | Not measured | Not in index |
| PostTrainBenchAgentic work | Not measured | Not in index |
| GPQA Diamond (Epoch AI run)Reasoning | Not measured | Not in index |
| ARC-AGIReasoning | Not measured | Not in index |
| ARC-AGI-2Reasoning | Not measured | Not in index |
| Humanity's Last ExamReasoning | Not measured | Not in index |
| SimpleBenchReasoning | Not measured | Not in index |
| Chess puzzlesReasoning | Not measured | Not in index |
| Mystery game puzzlesReasoning | Not measured | Not in index |
| EBR-benchReasoning | Not measured | Not in index |
| OTIS Mock AIMEMath | Not measured | Not in index |
| FrontierMath, tiers 1 to 3Math | Not measured | Not in index |
| FrontierMath, tier 4Math | Not measured | Not in index |
| ProofBenchMath | 42.0% | Not in index |
| MATH level 5Math | Not measured | Not in index |
| GSM8KMath | Not measured | Not in index |
| SimpleQA VerifiedKnowledge | Not measured | Not in index |
| MMLUKnowledge | Not measured | Not in index |
| GeoBenchKnowledge | Not measured | Not in index |
| DTBenchGeneral | Not measured | Not in index |
| LMCAGeneral | Not measured | Not in index |
| Fiction.LiveBenchGeneral | Not measured | Not in index |
| Creative writing (Lech Mazur)General | Not measured | Not in index |
| CL-benchGeneral | Not measured | Not in index |
| Surface Evolver BenchGeneral | Not measured | Not in index |
| Furniture assemblyGeneral | Not measured | Not in index |
| ExploitBenchGeneral | Not measured | Not in index |
| CybenchCoding | Not measured | Not in index |
| TheAgentCompanyAgentic work | Not measured | Not in index |
| OSWorldAgentic work | Not measured | Not in index |
| SpireClimbAgentic work | Not measured | Not in index |
| VPCTReasoning | Not measured | Not in index |
| BIG-Bench HardReasoning | Not measured | Not in index |
| AI2 Reasoning ChallengeKnowledge | Not measured | Not in index |
| HellaSwagGeneral | Not measured | Not in index |
| WinoGrandeGeneral | Not measured | Not in index |
| PIQAGeneral | Not measured | Not in index |
| CL-bench LifeGeneral | Not measured | Not in index |
| ALE-BenchCoding | Not measured | Not in index |
| Vending-Bench 2Agentic work | Not measured | Not in index |
| Epoch Capabilities IndexIntelligence | Not measured | Not in index |
| LMArena TextGeneral | 1440 | Not in index |
| LMArena Text: codingCoding | 1485 | Not in index |
| LMArena Text: hard promptsReasoning | 1461 | Not in index |
| LMArena Text: mathMath | 1458 | Not in index |
| LMArena Text: instruction followingGeneral | 1438 | Not in index |
| LMArena Text: creative writingGeneral | 1409 | Not in index |
| LMArena VisionGeneral | 1265 | Not in index |
| LMArena Text: EnglishGeneral | 1447 | Not in index |
| LMArena Text: non-EnglishGeneral | 1428 | Not in index |
| LMArena Text: ChineseGeneral | 1500 | Not in index |
| LMArena Text: GermanGeneral | Not measured | Not in index |
| LMArena Text: JapaneseGeneral | 1383 | Not in index |
| LMArena Text: KoreanGeneral | Not measured | Not in index |
| LMArena Text: FrenchGeneral | 1438 | Not in index |
| LMArena Text: SpanishGeneral | 1441 | Not in index |
| LMArena Text: RussianGeneral | 1433 | Not in index |
| LMArena Text: PolishGeneral | Not measured | Not in index |
| LMArena Text: multi-turnGeneral | 1437 | Not in index |
| LMArena Text: longer queriesGeneral | 1449 | Not in index |
| LMArena Text: expertReasoning | 1485 | Not in index |
| LMArena Text: hard prompts in EnglishReasoning | 1454 | Not in index |
| LMArena Text: software and ITCoding | 1473 | Not in index |
| LMArena Text: mathematical workMath | 1460 | Not in index |
| LMArena Text: scienceKnowledge | 1454 | Not in index |
| LMArena Text: medicine and healthcareKnowledge | 1439 | Not in index |
| LMArena Text: legal and governmentKnowledge | 1437 | Not in index |
| LMArena Text: business and financeKnowledge | 1435 | Not in index |
| LMArena Text: writing and languageGeneral | 1425 | Not in index |
| LMArena Text: entertainment and mediaGeneral | 1412 | Not in index |
| LMArena WebDev: ReactCoding | 1583 | Not in index |
| LMArena WebDev: HTMLCoding | 1556 | Not in index |
| LMArena WebDev: from an imageCoding | 1550 | Not in index |
| LMArena Vision: EnglishGeneral | 1290 | Not in index |
| LMArena Vision: ChineseGeneral | Not measured | Not in index |
| LMArena Vision: reading textGeneral | 1282 | Not in index |
| LMArena Vision: diagramsReasoning | 1308 | Not in index |
| LMArena Vision: homeworkReasoning | 1265 | Not in index |
| LMArena Vision: humorGeneral | Not measured | Not in index |
| LMArena Vision: captioningGeneral | Not measured | Not in index |
| LMArena Vision: recognizing thingsGeneral | Not measured | Not in index |
| LMArena Vision: creative writingGeneral | Not measured | Not in index |
| LMArena SearchAgentic work | Not measured | Not in index |
| LMArena DocumentGeneral | Not measured | Not in index |
| LMArena Text to ImageImage generation | Not measured | Not in index |
| LMArena Image EditImage generation | Not measured | Not in index |
| Berkeley Function Calling Leaderboard (BFCL)Agentic work | Not measured | Not in index |
| Kagi LLM BenchmarkReasoning | Not measured | Not in index |
| BrowseComp-PlusAgentic work | Not measured | Not in index |
| SciCodeCoding | Not measured | Not in index |
| FACTS Benchmark SuiteKnowledge | Not measured | Not in index |
A list price says what a token costs. The cost per solved task says what a working answer costs, which is the number that decides.
Neither board that publishes what a SWE-bench Verified attempt costs, Vals AI and swebench.com, has run this model. We leave the number out rather than estimate it.
LiveBench, on its own coding tasks, publishes what one task cost this model: $1.095, at a score of 85.6%. LiveBench counts that as $1.28 per successful task. Different tasks from SWE-bench, so it is not our number and does not enter its ranking. LiveBench coding
Beside a result, many boards publish what the run cost, how long it took, the tokens it spent or the votes behind a rating. Each figure as its board printed it, one board per row.
| Board | Result | What the run cost | Also published |
|---|---|---|---|
| LiveBench coding | 85.6% | $1.0953 per task | Output tokens per task 43,804 |
| LiveBench reasoning | 83.8% | $0.0482 per task | Output tokens per task 17,443 |
| LiveBench data analysis | 63.5% | $0.0639 per task | Output tokens per task 22,293 |
| LiveBench research | 72.8% | $1.2289 per task | Output tokens per task 72,511 |
| LiveBench browser | 24.0% | $2.7917 per task | Output tokens per task 107,108 |
| LiveBench long agentic | 27.5% | $10.3756 per task | Output tokens per task 493,230 |
| LiveBench design | 76.9% | $0.7457 per task | Output tokens per task 71,830 |
| Terminal-Bench 4 (Vals AI) | 31.8% | $3.11 per test | Duration, as printed 1h 58m |
| Vibe Code Bench (Vals AI) | 70.4% | $5.86 per test | Duration, as printed 1h 31m |
| Code migration (Vals AI) | 39.0% | $7.61 per test | Duration, as printed 3h 34m |
| SREBench (Vals AI) | 0.4% | $3.07 per test | Duration, as printed 1h 34m |
| IOI (Vals AI) | 38.1% | $2.03 per test | Duration, as printed 1h 20m |
| SWE-bench Multimodal (Vals AI) | 40.8% | $0.61 per test | Duration, as printed 26m 06s |
|---|---|---|---|
| Finance Agent (Vals AI) | 50.7% | $0.62 per test | Duration, as printed 12m 28s |
| Tax Agent Bench (Vals AI) | 27.3% | $0.55 per test | Duration, as printed 21m 29s |
| Legal Research Bench (Vals AI) | 40.4% | $0.84 per test | Duration, as printed 36m 08s |
| Excel Modeling Benchmark (Vals AI) | 57.1% | $1.33 per test | Duration, as printed 43m 35s |
| BioMysteryBench (Vals AI) | 67.4% | $0.49 per test | Duration, as printed 21m 35s |
| Terminal-Bench-Science (Vals AI) | 2.9% | $4.25 per test | Duration, as printed 2h 30m |
| Vals Science Index | 29.6% | $1.78 per test | Duration, as printed 1h 07m |
| Public Benefits Bench (Vals AI) | 58.1% | $0.50 per test | Nothing else |
| MysteryMechanism (Vals AI) | 18.5% | $0.60 per test | Duration, as printed 29m 20s |
| ProofBench | 42.0% | $0.7146 per test | Time per proof 38m 34s |
| LMArena WebDev | 1567 Elo | Not published | Votes 4,728 · Interval (±) ±9.8 |
| LMArena Text | 1440 Elo | Not published | Votes 8,756 · Interval (±) ±6.9 |
| LMArena Text: coding | 1485 Elo | Not published | Votes 2,074 · Interval (±) ±12.9 |
| LMArena Text: hard prompts | 1461 Elo | Not published | Votes 5,502 · Interval (±) ±8.4 |
| LMArena Text: math | 1458 Elo | Not published | Votes 443 · Interval (±) ±26.6 |
| LMArena Text: instruction following | 1438 Elo | Not published | Votes 3,099 · Interval (±) ±10.9 |
| LMArena Text: creative writing | 1409 Elo | Not published | Votes 1,871 · Interval (±) ±14.4 |
| LMArena Vision | 1265 Elo | Not published | Votes 1,649 · Interval (±) ±15.0 |
| LMArena Text: English | 1447 Elo | Not published | Votes 3,106 · Interval (±) ±10.8 |
| LMArena Text: non-English | 1428 Elo | Not published | Votes 5,647 · Interval (±) ±8.3 |
| LMArena Text: Chinese | 1500 Elo | Not published | Votes 710 · Interval (±) ±21.7 |
| LMArena Text: Japanese | 1383 Elo | Not published | Votes 168 · Interval (±) ±45.4 |
| LMArena Text: French | 1438 Elo | Not published | Votes 388 · Interval (±) ±30.9 |
| LMArena Text: Spanish | 1441 Elo | Not published | Votes 318 · Interval (±) ±31.9 |
| LMArena Text: Russian | 1433 Elo | Not published | Votes 1,288 · Interval (±) ±16.1 |
| LMArena Text: multi-turn | 1437 Elo | Not published | Votes 1,231 · Interval (±) ±17.0 |
| LMArena Text: longer queries | 1449 Elo | Not published | Votes 3,705 · Interval (±) ±10.2 |
| LMArena Text: expert | 1485 Elo | Not published | Votes 874 · Interval (±) ±19.5 |
| LMArena Text: hard prompts in English | 1454 Elo | Not published | Votes 2,022 · Interval (±) ±13.2 |
| LMArena Text: software and IT | 1473 Elo | Not published | Votes 3,192 · Interval (±) ±10.6 |
| LMArena Text: mathematical work | 1460 Elo | Not published | Votes 502 · Interval (±) ±25.4 |
| LMArena Text: science | 1454 Elo | Not published | Votes 1,416 · Interval (±) ±15.6 |
| LMArena Text: medicine and healthcare | 1439 Elo | Not published | Votes 614 · Interval (±) ±23.9 |
| LMArena Text: legal and government | 1437 Elo | Not published | Votes 748 · Interval (±) ±21.3 |
| LMArena Text: business and finance | 1435 Elo | Not published | Votes 1,506 · Interval (±) ±15.3 |
| LMArena Text: writing and language | 1425 Elo | Not published | Votes 2,398 · Interval (±) ±12.6 |
| LMArena Text: entertainment and media | 1412 Elo | Not published | Votes 2,237 · Interval (±) ±13.1 |
| LMArena WebDev: React | 1583 Elo | Not published | Votes 3,019 · Interval (±) ±12.0 |
| LMArena WebDev: HTML | 1556 Elo | Not published | Votes 722 · Interval (±) ±23.8 |
| LMArena WebDev: from an image | 1550 Elo | Not published | Votes 3,300 · Interval (±) ±11.6 |
| LMArena Vision: English | 1290 Elo | Not published | Votes 513 · Interval (±) ±26.2 |
| LMArena Vision: reading text | 1282 Elo | Not published | Votes 1,102 · Interval (±) ±17.9 |
| LMArena Vision: diagrams | 1308 Elo | Not published | Votes 423 · Interval (±) ±29.0 |
| LMArena Vision: homework | 1265 Elo | Not published | Votes 233 · Interval (±) ±37.8 |