Devstral 2 2512 vs Llama 3.2 3B Instruct
Updated October 21, 2026. 1 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.
Decision reading
Neither model has a coding score yet, so the two cannot be ranked on coding.
Devstral 2 2512 has the edge where the context window is larger, 262K against 131K tokens.
- Shared results
- 1
- Devstral 2 2512 only
- 5
- Llama 3.2 3B Instruct only
- 33
- Like-for-like categories
- 1 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, six public coding benchmarks
No call
Neither has a coding score yet.
One side not measured
Intelligence
AA Intelligence Index
Devstral 2 2512
Devstral 2 2512 leads, 8.6 against 5.7.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only Devstral 2 2512 has a AA Agentic Index result, 2.3.
Directional only
Long documents
Listed context window
Devstral 2 2512
Devstral 2 2512 takes 262K tokens against 131K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Llama 3.2 3B Instruct
Devstral 2 2512 $0.0014, Llama 3.2 3B Instruct $0.0002 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only Devstral 2 2512 has the number, $1.19.
Directional only
At a glance
- Coding index
- Devstral 2 2512 not benchmarked, Llama 3.2 3B Instruct not benchmarked.
- Widest category
- Intelligence, 2.9 points apart on average over 1 shared benchmark.
- Widest single gap
- AA Intelligence Index: Devstral 2 2512 8.6 against Llama 3.2 3B Instruct 5.7, 2.9 points apart on a 0 to 100 scale.
- Output price
- Devstral 2 2512 $2.00 against Llama 3.2 3B Instruct $0.3325 per million output tokens, 6.0x apart.
- Context window
- Devstral 2 2512 takes 262K tokens, Llama 3.2 3B Instruct 131K.
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 | Llama 3.2 3B Instruct | Ties |
|---|---|---|---|---|
| Intelligence | 1 | 0 | ||
| All shared boards | 1 | 0 |
CodingDirectional onlySix of these boards feed the Vibe Coding Index7 boards, none shared
| Benchmark | Devstral 2 2512 | Llama 3.2 3B Instruct | Leader |
|---|---|---|---|
| SWE-bench Verified | 53.8% | Not measured | Only Devstral 2 2512 measured |
| AA Coding Index | 31.3 | Not measured | Only Devstral 2 2512 measured |
| LiveCodeBench (Vals AI) | 51.8% | Not measured | Only Devstral 2 2512 measured |
| SWE-bench Verified (Vals AI) | 62.8% | Not measured | Only Devstral 2 2512 measured |
| LiveCodeBench (Artificial Analysis run) | Not measured | 8.3% | Only Llama 3.2 3B Instruct measured |
| LMArena Text: coding | Not measured | 1098 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: software and IT | Not measured | 1103 | Only Llama 3.2 3B Instruct measured |
| Neither model has a result on the other 31 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Devstral 2 2512 | Llama 3.2 3B Instruct | Leader |
|---|---|---|---|
| AA Intelligence Index | 8.6 | 5.7 | Devstral 2 2512 |
| Neither model has a result on the other 1 board here | |||
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index3 boards, none shared
| Benchmark | Devstral 2 2512 | Llama 3.2 3B Instruct | Leader |
|---|---|---|---|
| AA Agentic Index | 2.3 | Not measured | Only Devstral 2 2512 measured |
| τ²-Bench Telecom | Not measured | 21.1% | Only Llama 3.2 3B Instruct measured |
| BALROG | Not measured | 10.1% | Only Llama 3.2 3B Instruct measured |
| Neither model has a result on the other 36 boards here | |||
ReasoningDirectional onlyPublic boards, shown as published, kept out of the index5 boards, none shared
| Benchmark | Devstral 2 2512 | Llama 3.2 3B Instruct | Leader |
|---|---|---|---|
| GPQA Diamond (Artificial Analysis run) | Not measured | 25.5% | Only Llama 3.2 3B Instruct measured |
| Humanity's Last Exam (Artificial Analysis run) | Not measured | 5.3% | Only Llama 3.2 3B Instruct measured |
| LMArena Text: hard prompts | Not measured | 1095 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: expert | Not measured | 1090 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: hard prompts in English | Not measured | 1141 | Only Llama 3.2 3B Instruct measured |
| Neither model has a result on the other 16 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index4 boards, none shared
| Benchmark | Devstral 2 2512 | Llama 3.2 3B Instruct | Leader |
|---|---|---|---|
| AA Math Index | Not measured | 3.3 | Only Llama 3.2 3B Instruct measured |
| AIME 2025 (Artificial Analysis run) | Not measured | 3.3% | Only Llama 3.2 3B Instruct measured |
| LMArena Text: math | Not measured | 1126 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: mathematical work | Not measured | 1128 | Only Llama 3.2 3B Instruct measured |
| Neither model has a result on the other 9 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index5 boards, none shared
| Benchmark | Devstral 2 2512 | Llama 3.2 3B Instruct | Leader |
|---|---|---|---|
| MMLU Pro (Artificial Analysis run) | Not measured | 34.7% | Only Llama 3.2 3B Instruct measured |
| LMArena Text: science | Not measured | 1106 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: medicine and healthcare | Not measured | 1104 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: legal and government | Not measured | 1115 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: business and finance | Not measured | 1074 | Only Llama 3.2 3B Instruct measured |
| Neither model has a result on the other 12 boards here | |||
Web designNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Devstral 2 2512 | Llama 3.2 3B Instruct | Leader |
|---|---|---|---|
| Neither model has a result on 9 boards here | |||
GeneralDirectional onlyPublic boards, shown as published, kept out of the index14 boards, none shared
| Benchmark | Devstral 2 2512 | Llama 3.2 3B Instruct | Leader |
|---|---|---|---|
| IFBench | Not measured | 26.2% | Only Llama 3.2 3B Instruct measured |
| Long context reasoning (AA-LCR) | Not measured | 4.3% | Only Llama 3.2 3B Instruct measured |
| LMArena Text | Not measured | 1110 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: instruction following | Not measured | 1089 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: creative writing | Not measured | 1094 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: English | Not measured | 1180 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: non-English | Not measured | 1019 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: Chinese | Not measured | 1017 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: German | Not measured | 1056 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: Russian | Not measured | 949 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: multi-turn | Not measured | 1105 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: longer queries | Not measured | 1100 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: writing and language | Not measured | 1096 | Only Llama 3.2 3B Instruct measured |
| LMArena Text: entertainment and media | Not measured | 1083 | Only Llama 3.2 3B Instruct measured |
| Neither model has a result on the other 32 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Devstral 2 2512 | Llama 3.2 3B Instruct | 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
Llama 3.2 3B Instruct costs less
Repository review
50K input + 3K output tokens
Llama 3.2 3B Instruct costs less
Agent loop
200K input + 10K output tokens
Llama 3.2 3B Instruct 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 | Meta Llama 3.2 3B Instruct |
|---|---|---|
| Maker | Mistral AI | Meta Llama |
| Weights | Open weights | Open weights |
| Released | December 9, 2025 | September 25, 2024 |
| Context window | 262K | 131K |
| Input $/Mtok | $0.40 | $0.0505 |
| Output $/Mtok | $2.00 | $0.3325 |
| Coding index | Not benchmarked | Not benchmarked |
| Independent measurements behind it | 1 | 0 |
| Cost per solved task | $1.19 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Devstral 2 2512 better than Llama 3.2 3B Instruct for coding?
Neither model has a coding score yet, so the two cannot be ranked on coding. Devstral 2 2512 has the edge where the context window is larger, 262K against 131K tokens.
Which is cheaper, Devstral 2 2512 or Llama 3.2 3B Instruct?
Devstral 2 2512 costs $0.40 per million input tokens and $2.00 per million output tokens; Llama 3.2 3B Instruct costs $0.0505 and $0.3325. Cost per solved task is known for Devstral 2 2512 only.
Which has the bigger context window, Devstral 2 2512 or Llama 3.2 3B Instruct?
Devstral 2 2512, with 262K tokens against 131K.
How many benchmarks is the Devstral 2 2512 vs Llama 3.2 3B Instruct comparison based on?
1 benchmarks have a result for both models as of October 21, 2026. Devstral 2 2512's coding index rests on 1 independent benchmark and Llama 3.2 3B Instruct's on 0. Where only one model has a result, the page says so instead of naming a winner.