Qwen3.8 27B vs Step 5 Preview
Updated October 21, 2026. 54 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.
Decision reading
Pick Qwen3.8 27B for coding: it scores 73.6 on the Vibe Coding Index against 71.4 for Step 5 Preview.
Pick Step 5 Preview when the context window is larger, 1.05M against 1M tokens, or when what matters is Vibe Code Bench (Vals AI) or Code migration (Vals AI), where it leads.
- Shared results
- 54
- Qwen3.8 27B only
- 31
- Step 5 Preview only
- 7
- Like-for-like categories
- 8 of 9
Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.
Which one for your work
Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.
Coding
Vibe Coding Index, four public coding benchmarks
Qwen3.8 27B
Qwen3.8 27B scores 73.6 against 71.4.
2 benchmarks behind each score
Intelligence
AA Intelligence Index
Step 5 Preview
Step 5 Preview leads, 43.7 against 33.7.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only Qwen3.8 27B has a AA Agentic Index result, 45.8.
Directional only
Reasoning
GPQA Diamond
No call
Only Qwen3.8 27B has a GPQA Diamond result, 88.9%.
Directional only
Knowledge
MMLU Pro
No call
Only Qwen3.8 27B has a MMLU Pro result, 84.3%.
Directional only
Long documents
Listed context window
Step 5 Preview
Step 5 Preview takes 1.05M tokens against 1M.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Qwen3.8 27B
Qwen3.8 27B $0.0007, Step 5 Preview $0.0024 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only Qwen3.8 27B has the number, $1.26.
Directional only
At a glance
- Coding index
- Qwen3.8 27B 73.6 from 2 benchmarks, Step 5 Preview 71.4 from 2 benchmarks.
- Widest category
- Intelligence, 10.0 points apart on average over 1 shared benchmark.
- Widest single gap
- ProofBench: Qwen3.8 27B 16.0% against Step 5 Preview 42.0%, 26.0 points apart on a 0 to 100 scale.
- Output price
- Qwen3.8 27B $1.35 against Step 5 Preview $2.70 per million output tokens, 2.0x apart.
- Context window
- Qwen3.8 27B takes 1M tokens, Step 5 Preview 1.05M.
Benchmark by benchmark
First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.
| Category | Qwen3.8 27B | Bars | Step 5 Preview | Ties |
|---|---|---|---|---|
| Coding | 7 | 4 | ||
| Intelligence | 0 | 1 | ||
| Agentic work | 4 | 4 | ||
| Reasoning | 2 | 5 | ||
| Math | 0 | 3 | ||
| Knowledge | 3 | 1 | ||
| Web design | 0 | 1 | ||
| General | 8 | 11 | ||
| All shared boards | 24 | 30 |
CodingLike-for-likeFour of these boards feed the Vibe Coding Index74
| Benchmark | Qwen3.8 27B | Step 5 Preview | Leader |
|---|---|---|---|
| LiveBench coding | 86.8% | 85.6% | Qwen3.8 27B |
| LMArena WebDev | 1593 | 1567 | Qwen3.8 27B |
| AA Coding Index | 68.1 | Not measured | Only Qwen3.8 27B measured |
| LiveCodeBench (Vals AI) | 84.0% | Not measured | Only Qwen3.8 27B measured |
| SWE-bench Verified (Vals AI) | 86.0% | Not measured | Only Qwen3.8 27B measured |
| Terminal-Bench 4 (Vals AI) | Not measured | 31.8% | Only Step 5 Preview measured |
| Vibe Code Bench (Vals AI) | 64.8% | 70.4% | Step 5 Preview |
| Code migration (Vals AI) | 14.2% | 39.0% | Step 5 Preview |
| ProgramBench (Vals AI) | 0.0% | Not measured | Only Qwen3.8 27B measured |
| IOI (Vals AI) | 39.1% | 38.1% | Qwen3.8 27B |
| Terminal-Bench 2.1 (Vals AI) | 58.4% | Not measured | Only Qwen3.8 27B measured |
| SWE-bench Multimodal (Vals AI) | Not measured | 40.8% | Only Step 5 Preview measured |
| SkillsBench (Vals AI) | 38.1% | Not measured | Only Qwen3.8 27B measured |
| CyberBench (Vals AI) | 67.8% | Not measured | Only Qwen3.8 27B measured |
| SciCode | 46.6% | 58.9% | Step 5 Preview |
| LMArena Text: coding | 1483 | 1485 | Step 5 Preview |
| LMArena Text: software and IT | 1478 | 1473 | Qwen3.8 27B |
| LMArena WebDev: React | 1602 | 1583 | Qwen3.8 27B |
| LMArena WebDev: HTML | 1570 | 1556 | Qwen3.8 27B |
| LMArena WebDev: from an image | 1577 | 1550 | Qwen3.8 27B |
| Neither model has a result on the other 18 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | Qwen3.8 27B | Step 5 Preview | Leader |
|---|---|---|---|
| AA Intelligence Index | 33.7 | 43.7 | Step 5 Preview |
| Epoch Capabilities Index | 151.7 | Not measured | Only Qwen3.8 27B measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index44
| Benchmark | Qwen3.8 27B | Step 5 Preview | Leader |
|---|---|---|---|
| AA Agentic Index | 45.8 | Not measured | Only Qwen3.8 27B measured |
| LiveBench research | 59.4% | 72.8% | Step 5 Preview |
| LiveBench browser | 32.0% | 24.0% | Qwen3.8 27B |
| LiveBench long agentic | 27.9% | 27.5% | Qwen3.8 27B |
| SREBench (Vals AI) | Not measured | 0.4% | Only Step 5 Preview measured |
| Finance Agent (Vals AI) | 48.5% | 50.7% | Step 5 Preview |
| Tax Agent Bench (Vals AI) | 30.9% | 27.3% | Qwen3.8 27B |
| Legal Research Bench (Vals AI) | 36.1% | 40.4% | Step 5 Preview |
| Excel Modeling Benchmark (Vals AI) | 59.7% | 57.1% | Qwen3.8 27B |
| BioMysteryBench (Vals AI) | Not measured | 67.4% | Only Step 5 Preview measured |
| Terminal-Bench-Science (Vals AI) | 0.0% | 2.9% | Step 5 Preview |
| Vals Science Index | Not measured | 29.6% | Only Step 5 Preview measured |
| APEX-Agents | 47.5% | Not measured | Only Qwen3.8 27B measured |
| Neither model has a result on the other 26 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index25
| Benchmark | Qwen3.8 27B | Step 5 Preview | Leader |
|---|---|---|---|
| GPQA Diamond | 88.9% | Not measured | Only Qwen3.8 27B measured |
| LiveBench reasoning | 80.0% | 83.8% | Step 5 Preview |
| MysteryMechanism (Vals AI) | Not measured | 18.5% | Only Step 5 Preview measured |
| MMMU Pro (Vals AI) | 83.9% | Not measured | Only Qwen3.8 27B measured |
| GPQA Diamond (Artificial Analysis run) | 90.5% | Not measured | Only Qwen3.8 27B measured |
| Humanity's Last Exam (Artificial Analysis run) | 33.9% | 46.5% | Step 5 Preview |
| ARC-AGI | 87.5% | Not measured | Only Qwen3.8 27B measured |
| ARC-AGI-2 | 42.4% | Not measured | Only Qwen3.8 27B measured |
| LMArena Text: hard prompts | 1459 | 1461 | Step 5 Preview |
| LMArena Text: expert | 1479 | 1485 | Step 5 Preview |
| LMArena Text: hard prompts in English | 1461 | 1454 | Qwen3.8 27B |
| LMArena Vision: diagrams | 1285 | 1308 | Step 5 Preview |
| LMArena Vision: homework | 1295 | 1265 | Qwen3.8 27B |
| Neither model has a result on the other 8 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index03
| Benchmark | Qwen3.8 27B | Step 5 Preview | Leader |
|---|---|---|---|
| ProofBench | 16.0% | 42.0% | Step 5 Preview |
| LMArena Text: math | 1455 | 1458 | Step 5 Preview |
| LMArena Text: mathematical work | 1458 | 1460 | Step 5 Preview |
| Neither model has a result on the other 10 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index31
| Benchmark | Qwen3.8 27B | Step 5 Preview | Leader |
|---|---|---|---|
| MMLU Pro | 84.3% | Not measured | Only Qwen3.8 27B measured |
| LegalBench (Vals AI) | 82.4% | Not measured | Only Qwen3.8 27B measured |
| TaxEval (Vals AI) | 70.8% | Not measured | Only Qwen3.8 27B measured |
| MortgageTax (Vals AI) | 64.9% | Not measured | Only Qwen3.8 27B measured |
| MedCode (Vals AI) | 28.7% | Not measured | Only Qwen3.8 27B measured |
| LMArena Text: science | 1467 | 1454 | Qwen3.8 27B |
| LMArena Text: medicine and healthcare | 1437 | 1439 | Step 5 Preview |
| LMArena Text: legal and government | 1447 | 1437 | Qwen3.8 27B |
| LMArena Text: business and finance | 1438 | 1435 | Qwen3.8 27B |
| Neither model has a result on the other 8 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | Qwen3.8 27B | Step 5 Preview | Leader |
|---|---|---|---|
| LiveBench design | 75.9% | 76.9% | Step 5 Preview |
| Neither model has a result on the other 8 boards here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index811
| Benchmark | Qwen3.8 27B | Step 5 Preview | Leader |
|---|---|---|---|
| LiveBench data analysis | 76.6% | 63.5% | Qwen3.8 27B |
| Public Benefits Bench (Vals AI) | Not measured | 58.1% | Only Step 5 Preview measured |
| SAGE (Vals AI) | 52.4% | Not measured | Only Qwen3.8 27B measured |
| MedScribe (Vals AI) | 83.8% | Not measured | Only Qwen3.8 27B measured |
| Long context reasoning (AA-LCR) | 82.0% | 88.3% | Step 5 Preview |
| DTBench | 88.0% | Not measured | Only Qwen3.8 27B measured |
| LMCA | 41.4% | Not measured | Only Qwen3.8 27B measured |
| Surface Evolver Bench | 45.0% | Not measured | Only Qwen3.8 27B measured |
| LMArena Text | 1440 | 1440 | Step 5 Preview |
| LMArena Text: instruction following | 1436 | 1438 | Step 5 Preview |
| LMArena Text: creative writing | 1383 | 1409 | Step 5 Preview |
| LMArena Vision | 1269 | 1265 | Qwen3.8 27B |
| LMArena Text: English | 1448 | 1447 | Qwen3.8 27B |
| LMArena Text: non-English | 1428 | 1428 | Step 5 Preview |
| LMArena Text: Chinese | 1501 | 1500 | Qwen3.8 27B |
| LMArena Text: German | 1443 | Not measured | Only Qwen3.8 27B measured |
| LMArena Text: Japanese | 1387 | 1383 | Qwen3.8 27B |
| LMArena Text: Korean | 1384 | Not measured | Only Qwen3.8 27B measured |
| LMArena Text: French | 1465 | 1438 | Qwen3.8 27B |
| LMArena Text: Spanish | 1449 | 1441 | Qwen3.8 27B |
| LMArena Text: Russian | 1415 | 1433 | Step 5 Preview |
| LMArena Text: Polish | 1439 | Not measured | Only Qwen3.8 27B measured |
| LMArena Text: multi-turn | 1440 | 1437 | Qwen3.8 27B |
| LMArena Text: longer queries | 1448 | 1449 | Step 5 Preview |
| LMArena Text: writing and language | 1405 | 1425 | Step 5 Preview |
| LMArena Text: entertainment and media | 1393 | 1412 | Step 5 Preview |
| LMArena Vision: English | 1267 | 1290 | Step 5 Preview |
| LMArena Vision: Chinese | 1326 | Not measured | Only Qwen3.8 27B measured |
| LMArena Vision: reading text | 1281 | 1282 | Step 5 Preview |
| LMArena Vision: humor | 1235 | Not measured | Only Qwen3.8 27B measured |
| LMArena Vision: creative writing | 1237 | Not measured | Only Qwen3.8 27B measured |
| Neither model has a result on the other 15 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Qwen3.8 27B | Step 5 Preview | Leader |
|---|---|---|---|
| Neither model has a result on 5 boards here | |||
What each workload costs
Three fixed token mixes at each model's listed price, so per-token rates become a bill.
Chat turn
1K input + 500 output tokens
Qwen3.8 27B costs less
Repository review
50K input + 3K output tokens
Qwen3.8 27B costs less
Agent loop
200K input + 10K output tokens
Qwen3.8 27B costs less
Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.
Specification differences
The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.
| Specification | Qwen3.8 27B | StepFun Step 5 Preview |
|---|---|---|
| Maker | Qwen | StepFun |
| Weights | Open weights | Open weights |
| Released | August 14, 2026 | October 8, 2026 |
| Context window | 1M | 1.05M |
| Input $/Mtok | $0.04 | $1.00 |
| Output $/Mtok | $1.35 | $2.70 |
| Coding index | 73.6 | 71.4 |
| Independent measurements behind it | 2 | 2 |
| Cost per solved task | $1.26 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Qwen3.8 27B better than Step 5 Preview for coding?
Pick Qwen3.8 27B for coding: it scores 73.6 on the Vibe Coding Index against 71.4 for Step 5 Preview. Pick Step 5 Preview when the context window is larger, 1.05M against 1M tokens, or when what matters is Vibe Code Bench (Vals AI) or Code migration (Vals AI), where it leads.
Which is cheaper, Qwen3.8 27B or Step 5 Preview?
Qwen3.8 27B costs $0.04 per million input tokens and $1.35 per million output tokens; Step 5 Preview costs $1.00 and $2.70. Cost per solved task is known for Qwen3.8 27B only.
Which has the bigger context window, Qwen3.8 27B or Step 5 Preview?
Step 5 Preview, with 1.05M tokens against 1M.
How many benchmarks is the Qwen3.8 27B vs Step 5 Preview comparison based on?
54 benchmarks have a result for both models as of October 21, 2026. Qwen3.8 27B's coding index rests on 2 independent benchmarks and Step 5 Preview's on 2. Where only one model has a result, the page says so instead of naming a winner.