Devstral 2 2512 vs Qwen3.8 27B
Qwen3.8 27B leads Devstral 2 2512 on the Artificial Analysis Intelligence Index, 33.7 to 8.6, and Devstral 2 2512 costs 1.3x less per output token ($2.00 vs $2.55 per million). Updated October 21, 2026; 5 benchmarks have a result for both.
Devstral 2 2512 or Qwen3.8 27B: which to pick
Only Qwen3.8 27B has a coding score, 74.2 on the Vibe Coding Index; Devstral 2 2512 has none, so the two cannot be ranked on coding.
Pick Devstral 2 2512 for output tokens 1.3x cheaper ($2.00 vs $2.55 per million) and a lower cost per solved task ($1.19 against $1.26).
- Shared results
- 5
- Devstral 2 2512 only
- 1
- Qwen3.8 27B only
- 80
- Like-for-like categories
- 3 of 9
Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.
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, six public coding benchmarks
No call
Only Qwen3.8 27B has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
Qwen3.8 27B
Qwen3.8 27B leads, 33.7 against 8.6.
Same board, both measured
Agentic work
AA Agentic Index
Qwen3.8 27B
Qwen3.8 27B leads, 45.8 against 2.3.
Same board, both measured
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
Qwen3.8 27B
Qwen3.8 27B takes 1M tokens against 262K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Devstral 2 2512
Devstral 2 2512 $0.0014, Qwen3.8 27B $0.0017 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
Devstral 2 2512
Devstral 2 2512 $1.19, Qwen3.8 27B $1.26.
Published attempt cost and pass rate
At a glance
- Coding index
- Devstral 2 2512 not benchmarked, Qwen3.8 27B 74.2 from 4 benchmarks.
- Widest category
- Agentic work, 43.5 points apart on average over 1 shared benchmark.
- Widest single gap
- AA Agentic Index: Devstral 2 2512 2.3 against Qwen3.8 27B 45.8, 43.5 points apart on a 0 to 100 scale.
- Output price
- Devstral 2 2512 $2.00 against Qwen3.8 27B $2.55 per million output tokens, 1.3x apart.
- Context window
- Devstral 2 2512 takes 262K tokens, Qwen3.8 27B 1M.
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 | Devstral 2 2512 | Bars | Qwen3.8 27B | Ties |
|---|---|---|---|---|
| Coding | 0 | 3 | ||
| Intelligence | 0 | 1 | ||
| Agentic work | 0 | 1 | ||
| All shared boards | 0 | 5 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index03
| Benchmark | Devstral 2 2512 | Qwen3.8 27B | Leader |
|---|---|---|---|
| SWE-bench Verified | 53.8% | Not measured | Only Devstral 2 2512 measured |
| LiveBench coding | Not measured | 86.8% | Only Qwen3.8 27B measured |
| LMArena WebDev | Not measured | 1593 | Only Qwen3.8 27B measured |
| AA Coding Index | 31.3 | 68.1 | Qwen3.8 27B |
| LiveCodeBench (Vals AI) | 51.8% | 84.0% | Qwen3.8 27B |
| SWE-bench Verified (Vals AI) | 62.8% | 86.0% | Qwen3.8 27B |
| Vibe Code Bench (Vals AI) | Not measured | 64.8% | Only Qwen3.8 27B measured |
| Code migration (Vals AI) | Not measured | 14.2% | Only Qwen3.8 27B measured |
| ProgramBench (Vals AI) | Not measured | 0.0% | Only Qwen3.8 27B measured |
| IOI (Vals AI) | Not measured | 39.1% | Only Qwen3.8 27B measured |
| Terminal-Bench 2.1 (Vals AI) | Not measured | 58.4% | Only Qwen3.8 27B measured |
| SkillsBench (Vals AI) | Not measured | 38.1% | Only Qwen3.8 27B measured |
| CyberBench (Vals AI) | Not measured | 67.8% | Only Qwen3.8 27B measured |
| SciCode | Not measured | 46.6% | Only Qwen3.8 27B measured |
| LMArena Text: coding | Not measured | 1483 | Only Qwen3.8 27B measured |
| LMArena Text: software and IT | Not measured | 1478 | Only Qwen3.8 27B measured |
| LMArena WebDev: React | Not measured | 1602 | Only Qwen3.8 27B measured |
| LMArena WebDev: HTML | Not measured | 1570 | Only Qwen3.8 27B measured |
| LMArena WebDev: from an image | Not measured | 1577 | Only Qwen3.8 27B measured |
| Neither model has a result on the other 20 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | Devstral 2 2512 | Qwen3.8 27B | Leader |
|---|---|---|---|
| AA Intelligence Index | 8.6 | 33.7 | Qwen3.8 27B |
| Epoch Capabilities Index | Not measured | 151.7 | Only Qwen3.8 27B measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | Devstral 2 2512 | Qwen3.8 27B | Leader |
|---|---|---|---|
| AA Agentic Index | 2.3 | 45.8 | Qwen3.8 27B |
| LiveBench research | Not measured | 59.4% | Only Qwen3.8 27B measured |
| LiveBench browser | Not measured | 32.0% | Only Qwen3.8 27B measured |
| LiveBench long agentic | Not measured | 27.9% | Only Qwen3.8 27B measured |
| Finance Agent (Vals AI) | Not measured | 48.5% | Only Qwen3.8 27B measured |
| Tax Agent Bench (Vals AI) | Not measured | 30.9% | Only Qwen3.8 27B measured |
| Legal Research Bench (Vals AI) | Not measured | 36.1% | Only Qwen3.8 27B measured |
| Excel Modeling Benchmark (Vals AI) | Not measured | 59.7% | Only Qwen3.8 27B measured |
| Terminal-Bench-Science (Vals AI) | Not measured | 0.0% | Only Qwen3.8 27B measured |
| APEX-Agents | Not measured | 47.5% | Only Qwen3.8 27B measured |
| Neither model has a result on the other 31 boards here | |||
ReasoningDirectional onlyPublic boards, shown as published, kept out of the index12 boards, none shared
| Benchmark | Devstral 2 2512 | Qwen3.8 27B | Leader |
|---|---|---|---|
| GPQA Diamond | Not measured | 88.9% | Only Qwen3.8 27B measured |
| LiveBench reasoning | Not measured | 80.0% | Only Qwen3.8 27B measured |
| MMMU Pro (Vals AI) | Not measured | 83.9% | Only Qwen3.8 27B measured |
| GPQA Diamond (Artificial Analysis run) | Not measured | 90.5% | Only Qwen3.8 27B measured |
| Humanity's Last Exam (Artificial Analysis run) | Not measured | 33.9% | Only Qwen3.8 27B measured |
| ARC-AGI | Not measured | 87.5% | Only Qwen3.8 27B measured |
| ARC-AGI-2 | Not measured | 42.4% | Only Qwen3.8 27B measured |
| LMArena Text: hard prompts | Not measured | 1459 | Only Qwen3.8 27B measured |
| LMArena Text: expert | Not measured | 1479 | Only Qwen3.8 27B measured |
| LMArena Text: hard prompts in English | Not measured | 1461 | Only Qwen3.8 27B measured |
| LMArena Vision: diagrams | Not measured | 1285 | Only Qwen3.8 27B measured |
| LMArena Vision: homework | Not measured | 1295 | Only Qwen3.8 27B measured |
| Neither model has a result on the other 10 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index3 boards, none shared
| Benchmark | Devstral 2 2512 | Qwen3.8 27B | Leader |
|---|---|---|---|
| ProofBench | Not measured | 16.0% | Only Qwen3.8 27B measured |
| LMArena Text: math | Not measured | 1455 | Only Qwen3.8 27B measured |
| LMArena Text: mathematical work | Not measured | 1458 | Only Qwen3.8 27B measured |
| Neither model has a result on the other 10 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index9 boards, none shared
| Benchmark | Devstral 2 2512 | Qwen3.8 27B | Leader |
|---|---|---|---|
| MMLU Pro | Not measured | 84.3% | Only Qwen3.8 27B measured |
| LegalBench (Vals AI) | Not measured | 82.4% | Only Qwen3.8 27B measured |
| TaxEval (Vals AI) | Not measured | 70.8% | Only Qwen3.8 27B measured |
| MortgageTax (Vals AI) | Not measured | 64.9% | Only Qwen3.8 27B measured |
| MedCode (Vals AI) | Not measured | 28.7% | Only Qwen3.8 27B measured |
| LMArena Text: science | Not measured | 1467 | Only Qwen3.8 27B measured |
| LMArena Text: medicine and healthcare | Not measured | 1437 | Only Qwen3.8 27B measured |
| LMArena Text: legal and government | Not measured | 1447 | Only Qwen3.8 27B measured |
| LMArena Text: business and finance | Not measured | 1438 | Only Qwen3.8 27B measured |
| Neither model has a result on the other 9 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index1 board, none shared
| Benchmark | Devstral 2 2512 | Qwen3.8 27B | Leader |
|---|---|---|---|
| LiveBench design | Not measured | 75.9% | Only Qwen3.8 27B measured |
| Neither model has a result on the other 8 boards here | |||
GeneralDirectional onlyPublic boards, shown as published, kept out of the index30 boards, none shared
| Benchmark | Devstral 2 2512 | Qwen3.8 27B | Leader |
|---|---|---|---|
| LiveBench data analysis | Not measured | 76.6% | Only Qwen3.8 27B measured |
| SAGE (Vals AI) | Not measured | 52.4% | Only Qwen3.8 27B measured |
| MedScribe (Vals AI) | Not measured | 83.8% | Only Qwen3.8 27B measured |
| Long context reasoning (AA-LCR) | Not measured | 82.0% | Only Qwen3.8 27B measured |
| DTBench | Not measured | 88.0% | Only Qwen3.8 27B measured |
| LMCA | Not measured | 41.4% | Only Qwen3.8 27B measured |
| Surface Evolver Bench | Not measured | 45.0% | Only Qwen3.8 27B measured |
| LMArena Text | Not measured | 1440 | Only Qwen3.8 27B measured |
| LMArena Text: instruction following | Not measured | 1436 | Only Qwen3.8 27B measured |
| LMArena Text: creative writing | Not measured | 1383 | Only Qwen3.8 27B measured |
| LMArena Vision | Not measured | 1269 | Only Qwen3.8 27B measured |
| LMArena Text: English | Not measured | 1448 | Only Qwen3.8 27B measured |
| LMArena Text: non-English | Not measured | 1428 | Only Qwen3.8 27B measured |
| LMArena Text: Chinese | Not measured | 1501 | Only Qwen3.8 27B measured |
| LMArena Text: German | Not measured | 1443 | Only Qwen3.8 27B measured |
| LMArena Text: Japanese | Not measured | 1387 | Only Qwen3.8 27B measured |
| LMArena Text: Korean | Not measured | 1384 | Only Qwen3.8 27B measured |
| LMArena Text: French | Not measured | 1465 | Only Qwen3.8 27B measured |
| LMArena Text: Spanish | Not measured | 1449 | Only Qwen3.8 27B measured |
| LMArena Text: Russian | Not measured | 1415 | Only Qwen3.8 27B measured |
| LMArena Text: Polish | Not measured | 1439 | Only Qwen3.8 27B measured |
| LMArena Text: multi-turn | Not measured | 1440 | Only Qwen3.8 27B measured |
| LMArena Text: longer queries | Not measured | 1448 | Only Qwen3.8 27B measured |
| LMArena Text: writing and language | Not measured | 1405 | Only Qwen3.8 27B measured |
| LMArena Text: entertainment and media | Not measured | 1393 | Only Qwen3.8 27B measured |
| LMArena Vision: English | Not measured | 1267 | Only Qwen3.8 27B measured |
| LMArena Vision: Chinese | Not measured | 1326 | Only Qwen3.8 27B measured |
| LMArena Vision: reading text | Not measured | 1281 | Only Qwen3.8 27B measured |
| LMArena Vision: humor | Not measured | 1235 | Only Qwen3.8 27B measured |
| LMArena Vision: creative writing | Not measured | 1237 | Only Qwen3.8 27B measured |
| Neither model has a result on the other 16 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Devstral 2 2512 | Qwen3.8 27B | 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
Devstral 2 2512 costs less
Repository review
50K input + 3K output tokens
Devstral 2 2512 costs less
Agent loop
200K input + 10K output tokens
Devstral 2 2512 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 | Mistral Devstral 2 2512 | Qwen3.8 27B |
|---|---|---|
| Maker | Mistral AI | Qwen |
| Weights | Open weights | Open weights |
| Released | December 9, 2025 | August 14, 2026 |
| Context window | 262K | 1M |
| Input $/Mtok | $0.40 | $0.425 |
| Output $/Mtok | $2.00 | $2.55 |
| Coding index | Not benchmarked | 74.2 |
| Independent measurements behind it | 1 | 4 |
| Cost per solved task | $1.19 | $1.26 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Devstral 2 2512 better than Qwen3.8 27B for coding?
Only Qwen3.8 27B has a coding score, 74.2 on the Vibe Coding Index; Devstral 2 2512 has none, so the two cannot be ranked on coding. Pick Devstral 2 2512 for output tokens 1.3x cheaper ($2.00 vs $2.55 per million) and a lower cost per solved task ($1.19 against $1.26).
Which is cheaper, Devstral 2 2512 or Qwen3.8 27B?
Devstral 2 2512 costs $0.40 per million input tokens and $2.00 per million output tokens; Qwen3.8 27B costs $0.425 and $2.55. Per solved SWE-bench Verified task, Devstral 2 2512 costs $1.19 and Qwen3.8 27B $1.26.
Which has the bigger context window, Devstral 2 2512 or Qwen3.8 27B?
Qwen3.8 27B, with 1M tokens against 262K.
How many benchmarks is the Devstral 2 2512 vs Qwen3.8 27B comparison based on?
5 benchmarks have a result for both models as of October 21, 2026. Devstral 2 2512's coding index rests on 1 independent benchmark and Qwen3.8 27B's on 4. Where only one model has a result, the page says so instead of naming a winner.