Enter opens the highlighted result. Esc closes.

MiniMax M2.5 vs Mistral Large 3 2512

MiniMax M2.5 leads Mistral Large 3 2512 on the Artificial Analysis Intelligence Index, 22.8 to 9.3, and MiniMax M2.5 costs 1.3x less per output token ($1.20 vs $1.50 per million). Updated October 22, 2026; 19 benchmarks have a result for both.

Replace MiniMax M2.5 Replace Mistral Large 3 2512
MiniMax M2.5MiniMax
Coding
37.2
Overall
37.3
Output $/M
$1.20
16
Mistral Large 3 2512Mistral AI
Coding
No data
Overall
No data
Output $/M
$1.50
3
vs boards won of 19 shared

MiniMax M2.5 or Mistral Large 3 2512: which to pick

Only MiniMax M2.5 has a coding score, 37.2 on the Vibe Coding Index; Mistral Large 3 2512 has none, so the two cannot be ranked on coding.

Pick Mistral Large 3 2512 for a larger context window (262K against 205K tokens) and work measured by CaseLaw (Vals AI) or CorpFin (Vals AI), where it leads.

Model AMiniMax M2.5MiniMax37.2/1003 independent benchmarks$0.62 per solved taskModel BMistral Large 3 2512Mistral AINot benchmarked1 independent benchmark$0.77 per solved task
Shared results
19
MiniMax M2.5 only
52
Mistral Large 3 2512 only
9
Like-for-like categories
6 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.

At a glance

Coding index
MiniMax M2.5 37.2 from 3 benchmarks, Mistral Large 3 2512 not benchmarked.
Widest category
Math, 45.8 points apart on average over 1 shared benchmark.
Widest single gap
AIME: MiniMax M2.5 88.8% against Mistral Large 3 2512 42.9%, 45.8 points apart on a 0 to 100 scale.
Output price
MiniMax M2.5 $1.20 against Mistral Large 3 2512 $1.50 per million output tokens, 1.3x apart.
Context window
MiniMax M2.5 takes 205K tokens, Mistral Large 3 2512 262K.

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.

Boards won, where both have a result
CategoryMiniMax M2.5BarsMistral Large 3 2512Ties
Coding30
Intelligence10
Reasoning10
Math10
Knowledge33
Web design70
All shared boards163
CodingLike-for-likeSix of these boards feed the Vibe Coding Index30
Coding benchmarks for MiniMax M2.5 and Mistral Large 3 2512
BenchmarkMiniMax M2.5Mistral Large 3 2512Leader
SWE-bench Verified75.8%Not measuredOnly MiniMax M2.5 measured
SWE-bench Multilingual68.3%Not measuredOnly MiniMax M2.5 measured
LMArena WebDev1386Not measuredOnly MiniMax M2.5 measured
AA Coding IndexNot measured20.1Only Mistral Large 3 2512 measured
LiveCodeBench (Vals AI)79.2%55.3%MiniMax M2.5
SWE-bench Verified (Vals AI)74.2%41.4%MiniMax M2.5
Vibe Code Bench (Vals AI)14.8%Not measuredOnly MiniMax M2.5 measured
Terminal-Bench Hard (Artificial Analysis run)34.8%Not measuredOnly MiniMax M2.5 measured
Terminal-Bench42.7%Not measuredOnly MiniMax M2.5 measured
ALE-Bench618.2264.7MiniMax M2.5
LMArena Text: coding1380Not measuredOnly MiniMax M2.5 measured
LMArena Text: software and IT1383Not measuredOnly MiniMax M2.5 measured
LMArena WebDev: React1374Not measuredOnly MiniMax M2.5 measured
LMArena WebDev: HTML1414Not measuredOnly MiniMax M2.5 measured
SWE Atlas: Codebase QnA10.3%Not measuredOnly MiniMax M2.5 measured
SWE Atlas: Refactoring19.5%Not measuredOnly MiniMax M2.5 measured
SWE Atlas: Test Writing18.6%Not measuredOnly MiniMax M2.5 measured
Neither model has a result on the other 26 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
Intelligence benchmarks for MiniMax M2.5 and Mistral Large 3 2512
BenchmarkMiniMax M2.5Mistral Large 3 2512Leader
AA Intelligence Index22.89.3MiniMax M2.5
Epoch Capabilities Index146.6Not measuredOnly MiniMax M2.5 measured
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index3 boards, none shared
Agentic work benchmarks for MiniMax M2.5 and Mistral Large 3 2512
BenchmarkMiniMax M2.5Mistral Large 3 2512Leader
AA Agentic IndexNot measured1.0Only Mistral Large 3 2512 measured
τ²-Bench Telecom95.3%Not measuredOnly MiniMax M2.5 measured
Vending-Bench 2-$23Not measuredOnly MiniMax M2.5 measured
Neither model has a result on the other 39 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index10
Reasoning benchmarks for MiniMax M2.5 and Mistral Large 3 2512
BenchmarkMiniMax M2.5Mistral Large 3 2512Leader
GPQA Diamond82.1%68.4%MiniMax M2.5
MMMU Pro (Vals AI)Not measured66.2%Only Mistral Large 3 2512 measured
GPQA Diamond (Artificial Analysis run)84.8%Not measuredOnly MiniMax M2.5 measured
Humanity's Last Exam (Artificial Analysis run)20.5%Not measuredOnly MiniMax M2.5 measured
ARC-AGI63.7%Not measuredOnly MiniMax M2.5 measured
ARC-AGI-24.9%Not measuredOnly MiniMax M2.5 measured
LMArena Text: hard prompts1372Not measuredOnly MiniMax M2.5 measured
LMArena Text: expert1378Not measuredOnly MiniMax M2.5 measured
LMArena Text: hard prompts in English1384Not measuredOnly MiniMax M2.5 measured
Kagi LLM Benchmark55.2%Not measuredOnly MiniMax M2.5 measured
Neither model has a result on the other 13 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index10
Math benchmarks for MiniMax M2.5 and Mistral Large 3 2512
BenchmarkMiniMax M2.5Mistral Large 3 2512Leader
AIME88.8%42.9%MiniMax M2.5
MGSM (Vals AI)Not measured85.4%Only Mistral Large 3 2512 measured
ProofBench4.0%Not measuredOnly MiniMax M2.5 measured
LMArena Text: math1378Not measuredOnly MiniMax M2.5 measured
LMArena Text: mathematical work1378Not measuredOnly MiniMax M2.5 measured
Neither model has a result on the other 9 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index33
Knowledge benchmarks for MiniMax M2.5 and Mistral Large 3 2512
BenchmarkMiniMax M2.5Mistral Large 3 2512Leader
MMLU Pro80.1%79.8%MiniMax M2.5
LegalBench (Vals AI)80.0%79.1%MiniMax M2.5
CaseLaw (Vals AI)53.5%61.4%Mistral Large 3 2512
CorpFin (Vals AI)59.6%61.0%Mistral Large 3 2512
TaxEval (Vals AI)68.2%73.1%Mistral Large 3 2512
MortgageTax (Vals AI)Not measured52.1%Only Mistral Large 3 2512 measured
MedQA (Vals AI)92.5%82.2%MiniMax M2.5
LMArena Text: science1369Not measuredOnly MiniMax M2.5 measured
LMArena Text: medicine and healthcare1357Not measuredOnly MiniMax M2.5 measured
LMArena Text: legal and government1366Not measuredOnly MiniMax M2.5 measured
LMArena Text: business and finance1356Not measuredOnly MiniMax M2.5 measured
Neither model has a result on the other 9 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index70
Web design benchmarks for MiniMax M2.5 and Mistral Large 3 2512
BenchmarkMiniMax M2.5Mistral Large 3 2512Leader
Design Arena: websites12221162MiniMax M2.5
Design Arena: UI components11701099MiniMax M2.5
Design Arena: game dev11761079MiniMax M2.5
Design Arena: data viz11721137MiniMax M2.5
Design Arena: 3D11701099MiniMax M2.5
Design Arena: all code categories12071141MiniMax M2.5
Design Arena: SVG1150992MiniMax M2.5
Design Arena: ASCII artNot measured1069Only Mistral Large 3 2512 measured
Neither model has a result on the other 1 board here
GeneralDirectional onlyPublic boards, shown as published, kept out of the index24 boards, none shared
General benchmarks for MiniMax M2.5 and Mistral Large 3 2512
BenchmarkMiniMax M2.5Mistral Large 3 2512Leader
SAGE (Vals AI)Not measured24.6%Only Mistral Large 3 2512 measured
IFBench71.6%Not measuredOnly MiniMax M2.5 measured
Long context reasoning (AA-LCR)73.3%Not measuredOnly MiniMax M2.5 measured
DTBenchNot measured65.1%Only Mistral Large 3 2512 measured
LMCANot measured16.7%Only Mistral Large 3 2512 measured
CL-bench11.4%Not measuredOnly MiniMax M2.5 measured
CL-bench Life6.3%Not measuredOnly MiniMax M2.5 measured
LMArena Text1359Not measuredOnly MiniMax M2.5 measured
LMArena Text: instruction following1353Not measuredOnly MiniMax M2.5 measured
LMArena Text: creative writing1331Not measuredOnly MiniMax M2.5 measured
LMArena Text: English1380Not measuredOnly MiniMax M2.5 measured
LMArena Text: non-English1338Not measuredOnly MiniMax M2.5 measured
LMArena Text: Chinese1393Not measuredOnly MiniMax M2.5 measured
LMArena Text: German1361Not measuredOnly MiniMax M2.5 measured
LMArena Text: Japanese1171Not measuredOnly MiniMax M2.5 measured
LMArena Text: Korean1233Not measuredOnly MiniMax M2.5 measured
LMArena Text: French1363Not measuredOnly MiniMax M2.5 measured
LMArena Text: Spanish1353Not measuredOnly MiniMax M2.5 measured
LMArena Text: Russian1359Not measuredOnly MiniMax M2.5 measured
LMArena Text: Polish1334Not measuredOnly MiniMax M2.5 measured
LMArena Text: multi-turn1364Not measuredOnly MiniMax M2.5 measured
LMArena Text: longer queries1366Not measuredOnly MiniMax M2.5 measured
LMArena Text: writing and language1345Not measuredOnly MiniMax M2.5 measured
LMArena Text: entertainment and media1329Not measuredOnly MiniMax M2.5 measured
Neither model has a result on the other 24 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for MiniMax M2.5 and Mistral Large 3 2512
BenchmarkMiniMax M2.5Mistral Large 3 2512Leader
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

MiniMax M2.5$0.0009Fits in one request
Mistral Large 3 2512$0.0013Fits in one request

MiniMax M2.5 costs less

Repository review

50K input + 3K output tokens

MiniMax M2.5$0.019Fits in one request
Mistral Large 3 2512$0.029Fits in one request

MiniMax M2.5 costs less

Agent loop

200K input + 10K output tokens

MiniMax M2.5$0.072Exceeds the context window
Mistral Large 3 2512$0.115Fits in one request

MiniMax M2.5 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.

Specifications of MiniMax M2.5 and Mistral Large 3 2512
SpecificationMiniMax M2.5Mistral Large 3 2512
MakerMiniMaxMistral AI
WeightsOpen weightsNo public weights listed
ReleasedFebruary 12, 2026December 1, 2025
Context window205K262K
Input $/Mtok$0.30$0.50
Output $/Mtok$1.20$1.50
Coding index37.2Not benchmarked
Independent measurements behind it31
Cost per solved task$0.62$0.77
SourceOpenRouter listingOpenRouter listing

Questions

Is MiniMax M2.5 better than Mistral Large 3 2512 for coding?

Only MiniMax M2.5 has a coding score, 37.2 on the Vibe Coding Index; Mistral Large 3 2512 has none, so the two cannot be ranked on coding. Pick Mistral Large 3 2512 for a larger context window (262K against 205K tokens) and work measured by CaseLaw (Vals AI) or CorpFin (Vals AI), where it leads.

Which is cheaper, MiniMax M2.5 or Mistral Large 3 2512?

MiniMax M2.5 costs $0.30 per million input tokens and $1.20 per million output tokens; Mistral Large 3 2512 costs $0.50 and $1.50. Per solved SWE-bench Verified task, MiniMax M2.5 costs $0.62 and Mistral Large 3 2512 $0.77.

Which has the bigger context window, MiniMax M2.5 or Mistral Large 3 2512?

Mistral Large 3 2512, with 262K tokens against 205K.

How many benchmarks is the MiniMax M2.5 vs Mistral Large 3 2512 comparison based on?

19 benchmarks have a result for both models as of October 22, 2026. MiniMax M2.5's coding index rests on 3 independent benchmarks and Mistral Large 3 2512's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next