Devstral 2 2512 vs Gemini 3.1 Flash Lite Preview
Updated October 21, 2026. 5 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.
Decision reading
Pick Devstral 2 2512 for coding: it scores 53.8 on the Vibe Coding Index against 8.1 for Gemini 3.1 Flash Lite Preview.
Pick Gemini 3.1 Flash Lite Preview when output tokens cost 1.3x less, or when a solved task costs $0.18 against $1.19.
- Shared results
- 5
- Devstral 2 2512 only
- 1
- Gemini 3.1 Flash Lite Preview only
- 72
- Like-for-like categories
- 3 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
Devstral 2 2512
Devstral 2 2512 scores 53.8 against 8.1.
One benchmark behind each score
Intelligence
AA Intelligence Index
Gemini 3.1 Flash Lite Preview
Gemini 3.1 Flash Lite Preview leads, 15.6 against 8.6.
Same board, both measured
Agentic work
AA Agentic Index
Devstral 2 2512
Devstral 2 2512 leads, 2.3 against 1.6.
Same board, both measured
Reasoning
GPQA Diamond
No call
Only Gemini 3.1 Flash Lite Preview has a GPQA Diamond result, 81.1%.
Directional only
Math
AIME
No call
Only Gemini 3.1 Flash Lite Preview has a AIME result, 83.3%.
Directional only
Knowledge
MMLU Pro
No call
Only Gemini 3.1 Flash Lite Preview has a MMLU Pro result, 86.2%.
Directional only
Web design
Design Arena: websites
No call
Only Gemini 3.1 Flash Lite Preview has a Design Arena: websites result, 1083.
Directional only
Long documents
Listed context window
Gemini 3.1 Flash Lite Preview
Gemini 3.1 Flash Lite Preview takes 1.05M tokens against 262K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Gemini 3.1 Flash Lite Preview
Devstral 2 2512 $0.0014, Gemini 3.1 Flash Lite Preview $0.0010 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
Gemini 3.1 Flash Lite Preview
Devstral 2 2512 $1.19, Gemini 3.1 Flash Lite Preview $0.18.
Published attempt cost and pass rate
At a glance
- Coding index
- Devstral 2 2512 53.8 from 1 benchmark, Gemini 3.1 Flash Lite Preview 8.1 from 1 benchmark.
- Widest category
- Coding, 10.6 points apart on average over 3 shared benchmarks.
- Widest single gap
- LiveCodeBench (Vals AI): Devstral 2 2512 51.8% against Gemini 3.1 Flash Lite Preview 80.1%, 28.3 points apart on a 0 to 100 scale.
- Output price
- Devstral 2 2512 $2.00 against Gemini 3.1 Flash Lite Preview $1.50 per million output tokens, 1.3x apart.
- Context window
- Devstral 2 2512 takes 262K tokens, Gemini 3.1 Flash Lite 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 | Devstral 2 2512 | Bars | Gemini 3.1 Flash Lite Preview | Ties |
|---|---|---|---|---|
| Coding | 0 | 2 | 1 tie | |
| Intelligence | 0 | 1 | ||
| Agentic work | 1 | 0 | ||
| All shared boards | 1 | 3 | 1 tie |
CodingLike-for-likeFour of these boards feed the Vibe Coding Index02
| Benchmark | Devstral 2 2512 | Gemini 3.1 Flash Lite Preview | Leader |
|---|---|---|---|
| SWE-bench Verified | 53.8% | Not measured | Only Devstral 2 2512 measured |
| LMArena WebDev | Not measured | 1257 | Only Gemini 3.1 Flash Lite Preview measured |
| AA Coding Index | 31.3 | 34.7 | Gemini 3.1 Flash Lite Preview |
| LiveCodeBench (Vals AI) | 51.8% | 80.1% | Gemini 3.1 Flash Lite Preview |
| SWE-bench Verified (Vals AI) | 62.8% | 62.8% | Tie |
| Vibe Code Bench (Vals AI) | Not measured | 0.0% | Only Gemini 3.1 Flash Lite Preview measured |
| Code migration (Vals AI) | Not measured | 4.6% | Only Gemini 3.1 Flash Lite Preview measured |
| ProgramBench (Vals AI) | Not measured | 0.0% | Only Gemini 3.1 Flash Lite Preview measured |
| Terminal-Bench 2.1 (Vals AI) | Not measured | 34.1% | Only Gemini 3.1 Flash Lite Preview measured |
| ALE-Bench | Not measured | 797.7 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: coding | Not measured | 1400 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: software and IT | Not measured | 1413 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena WebDev: React | Not measured | 1245 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena WebDev: HTML | Not measured | 1276 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena WebDev: from an image | Not measured | 1336 | Only Gemini 3.1 Flash Lite Preview measured |
| Neither model has a result on the other 23 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | Devstral 2 2512 | Gemini 3.1 Flash Lite Preview | Leader |
|---|---|---|---|
| AA Intelligence Index | 8.6 | 15.6 | Gemini 3.1 Flash Lite Preview |
| Neither model has a result on the other 1 board here | |||
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Devstral 2 2512 | Gemini 3.1 Flash Lite Preview | Leader |
|---|---|---|---|
| AA Agentic Index | 2.3 | 1.6 | Devstral 2 2512 |
| Finance Agent (Vals AI) | Not measured | 30.0% | Only Gemini 3.1 Flash Lite Preview measured |
| Tax Agent Bench (Vals AI) | Not measured | 3.1% | Only Gemini 3.1 Flash Lite Preview measured |
| Legal Research Bench (Vals AI) | Not measured | 3.4% | Only Gemini 3.1 Flash Lite Preview measured |
| Excel Modeling Benchmark (Vals AI) | Not measured | 8.6% | Only Gemini 3.1 Flash Lite Preview measured |
| Neither model has a result on the other 34 boards here | |||
ReasoningDirectional onlyPublic boards, shown as published, kept out of the index7 boards, none shared
| Benchmark | Devstral 2 2512 | Gemini 3.1 Flash Lite Preview | Leader |
|---|---|---|---|
| GPQA Diamond | Not measured | 81.1% | Only Gemini 3.1 Flash Lite Preview measured |
| MMMU Pro (Vals AI) | Not measured | 82.5% | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: hard prompts | Not measured | 1407 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: expert | Not measured | 1398 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: hard prompts in English | Not measured | 1403 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Vision: diagrams | Not measured | 1242 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Vision: homework | Not measured | 1271 | Only Gemini 3.1 Flash Lite Preview measured |
| Neither model has a result on the other 14 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index3 boards, none shared
| Benchmark | Devstral 2 2512 | Gemini 3.1 Flash Lite Preview | Leader |
|---|---|---|---|
| AIME | Not measured | 83.3% | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: math | Not measured | 1428 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: mathematical work | Not measured | 1417 | Only Gemini 3.1 Flash Lite Preview measured |
| Neither model has a result on the other 10 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index11 boards, none shared
| Benchmark | Devstral 2 2512 | Gemini 3.1 Flash Lite Preview | Leader |
|---|---|---|---|
| MMLU Pro | Not measured | 86.2% | Only Gemini 3.1 Flash Lite Preview measured |
| LegalBench (Vals AI) | Not measured | 83.8% | Only Gemini 3.1 Flash Lite Preview measured |
| CaseLaw (Vals AI) | Not measured | 55.0% | Only Gemini 3.1 Flash Lite Preview measured |
| CorpFin (Vals AI) | Not measured | 59.4% | Only Gemini 3.1 Flash Lite Preview measured |
| TaxEval (Vals AI) | Not measured | 71.8% | Only Gemini 3.1 Flash Lite Preview measured |
| MortgageTax (Vals AI) | Not measured | 68.0% | Only Gemini 3.1 Flash Lite Preview measured |
| MedCode (Vals AI) | Not measured | 47.6% | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: science | Not measured | 1425 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: medicine and healthcare | Not measured | 1428 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: legal and government | Not measured | 1418 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: business and finance | Not measured | 1389 | Only Gemini 3.1 Flash Lite Preview measured |
| Neither model has a result on the other 6 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
| Benchmark | Devstral 2 2512 | Gemini 3.1 Flash Lite Preview | Leader |
|---|---|---|---|
| Design Arena: websites | Not measured | 1083 | Only Gemini 3.1 Flash Lite Preview measured |
| Design Arena: UI components | Not measured | 1071 | Only Gemini 3.1 Flash Lite Preview measured |
| Design Arena: game dev | Not measured | 1029 | Only Gemini 3.1 Flash Lite Preview measured |
| Design Arena: data viz | Not measured | 1045 | Only Gemini 3.1 Flash Lite Preview measured |
| Design Arena: 3D | Not measured | 1048 | Only Gemini 3.1 Flash Lite Preview measured |
| Design Arena: all code categories | Not measured | 1072 | Only Gemini 3.1 Flash Lite Preview measured |
| Design Arena: SVG | Not measured | 1052 | Only Gemini 3.1 Flash Lite Preview measured |
| Design Arena: ASCII art | Not measured | 1166 | Only Gemini 3.1 Flash Lite Preview measured |
| Neither model has a result on the other 1 board here | |||
GeneralDirectional onlyPublic boards, shown as published, kept out of the index28 boards, none shared
| Benchmark | Devstral 2 2512 | Gemini 3.1 Flash Lite Preview | Leader |
|---|---|---|---|
| SAGE (Vals AI) | Not measured | 49.5% | Only Gemini 3.1 Flash Lite Preview measured |
| MedScribe (Vals AI) | Not measured | 63.9% | Only Gemini 3.1 Flash Lite Preview measured |
| Vals Multimodal Index | Not measured | 41.4% | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text | Not measured | 1416 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: instruction following | Not measured | 1377 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: creative writing | Not measured | 1401 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Vision | Not measured | 1240 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: English | Not measured | 1414 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: non-English | Not measured | 1410 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: Chinese | Not measured | 1460 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: German | Not measured | 1430 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: Japanese | Not measured | 1413 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: Korean | Not measured | 1392 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: French | Not measured | 1424 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: Spanish | Not measured | 1421 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: Russian | Not measured | 1420 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: Polish | Not measured | 1443 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: multi-turn | Not measured | 1417 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: longer queries | Not measured | 1394 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: writing and language | Not measured | 1408 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Text: entertainment and media | Not measured | 1386 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Vision: English | Not measured | 1232 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Vision: Chinese | Not measured | 1278 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Vision: reading text | Not measured | 1246 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Vision: humor | Not measured | 1266 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Vision: captioning | Not measured | 1259 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Vision: recognizing things | Not measured | 1308 | Only Gemini 3.1 Flash Lite Preview measured |
| LMArena Vision: creative writing | Not measured | 1248 | Only Gemini 3.1 Flash Lite Preview measured |
| Neither model has a result on the other 18 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Devstral 2 2512 | Gemini 3.1 Flash Lite 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
Gemini 3.1 Flash Lite Preview costs less
Repository review
50K input + 3K output tokens
Gemini 3.1 Flash Lite Preview costs less
Agent loop
200K input + 10K output tokens
Gemini 3.1 Flash Lite Preview 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 | Google Gemini 3.1 Flash Lite Preview |
|---|---|---|
| Maker | Mistral AI | |
| Weights | Open weights | No public weights listed |
| Released | December 9, 2025 | March 3, 2026 |
| Context window | 262K | 1.05M |
| Input $/Mtok | $0.40 | $0.25 |
| Output $/Mtok | $2.00 | $1.50 |
| Coding index | 53.8 | 8.1 |
| Independent measurements behind it | 1 | 1 |
| Cost per solved task | $1.19 | $0.18 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Devstral 2 2512 better than Gemini 3.1 Flash Lite Preview for coding?
Pick Devstral 2 2512 for coding: it scores 53.8 on the Vibe Coding Index against 8.1 for Gemini 3.1 Flash Lite Preview. Pick Gemini 3.1 Flash Lite Preview when output tokens cost 1.3x less, or when a solved task costs $0.18 against $1.19.
Which is cheaper, Devstral 2 2512 or Gemini 3.1 Flash Lite Preview?
Devstral 2 2512 costs $0.40 per million input tokens and $2.00 per million output tokens; Gemini 3.1 Flash Lite Preview costs $0.25 and $1.50. Per solved SWE-bench Verified task, Devstral 2 2512 costs $1.19 and Gemini 3.1 Flash Lite Preview $0.18.
Which has the bigger context window, Devstral 2 2512 or Gemini 3.1 Flash Lite Preview?
Gemini 3.1 Flash Lite Preview, with 1.05M tokens against 262K.
How many benchmarks is the Devstral 2 2512 vs Gemini 3.1 Flash Lite Preview 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 Gemini 3.1 Flash Lite Preview's on 1. Where only one model has a result, the page says so instead of naming a winner.