Enter opens the highlighted result. Esc closes.

Mistral Medium 3.5 vs Mistral Small 3.1 24B

Mistral Small 3.1 24B costs 13.5x less per output token than Mistral Medium 3.5 ($0.555 vs $7.50 per million). Updated October 21, 2026; 35 benchmarks have a result for both.

Replace Mistral Medium 3.5 Replace Mistral Small 3.1 24B
Mistral Medium 3.5Mistral AI
Coding
25.5
Overall
37.3
Output $/M
$7.50
35
Mistral Small 3.1 24BMistral AI
Coding
No data
Overall
No data
Output $/M
$0.555
0
vs boards won of 35 shared

Mistral Medium 3.5 or Mistral Small 3.1 24B: which to pick

Only Mistral Medium 3.5 has a coding score, 25.5 on the Vibe Coding Index; Mistral Small 3.1 24B has none, so the two cannot be ranked on coding.

Pick Mistral Small 3.1 24B for output tokens 13.5x cheaper ($0.555 vs $7.50 per million).

Model AMistral Medium 3.5Mistral AI25.5/1003 independent benchmarks$6.02 per solved taskModel BMistral Small 3.1 24BMistral AINot benchmarkedNo coding benchmarkNo cost per solved task
Shared results
35
Mistral Medium 3.5 only
46
Mistral Small 3.1 24B only
3
Like-for-like categories
5 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
Mistral Medium 3.5 25.5 from 3 benchmarks, Mistral Small 3.1 24B not benchmarked.
Widest category
Math, 28.3 points apart on average over 2 shared benchmarks.
Widest single gap
LMArena Text: Chinese: Mistral Medium 3.5 1444 against Mistral Small 3.1 24B 1253, 31.9 points apart on a 0 to 100 scale.
Output price
Mistral Medium 3.5 $7.50 against Mistral Small 3.1 24B $0.555 per million output tokens, 13.5x apart.
Context window
Mistral Medium 3.5 takes 262K tokens, Mistral Small 3.1 24B 128K.

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
CategoryMistral Medium 3.5BarsMistral Small 3.1 24BTies
Coding20
Reasoning50
Math20
Knowledge40
General220
All shared boards350
CodingLike-for-likeSix of these boards feed the Vibe Coding Index20
Coding benchmarks for Mistral Medium 3.5 and Mistral Small 3.1 24B
BenchmarkMistral Medium 3.5Mistral Small 3.1 24BLeader
LMArena WebDev1264Not measuredOnly Mistral Medium 3.5 measured
AA Coding Index46.9Not measuredOnly Mistral Medium 3.5 measured
SWE-bench Verified (Vals AI)66.4%Not measuredOnly Mistral Medium 3.5 measured
Vibe Code Bench (Vals AI)2.9%Not measuredOnly Mistral Medium 3.5 measured
Code migration (Vals AI)5.1%Not measuredOnly Mistral Medium 3.5 measured
Terminal-Bench 2.1 (Vals AI)39.0%Not measuredOnly Mistral Medium 3.5 measured
SciCode40.2%Not measuredOnly Mistral Medium 3.5 measured
Terminal-Bench Hard (Artificial Analysis run)33.3%Not measuredOnly Mistral Medium 3.5 measured
ALE-Bench764.0Not measuredOnly Mistral Medium 3.5 measured
LMArena Text: coding14601309Mistral Medium 3.5
LMArena Text: software and IT14581301Mistral Medium 3.5
LMArena WebDev: React1254Not measuredOnly Mistral Medium 3.5 measured
LMArena WebDev: HTML1308Not measuredOnly Mistral Medium 3.5 measured
Neither model has a result on the other 26 boards here
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
Intelligence benchmarks for Mistral Medium 3.5 and Mistral Small 3.1 24B
BenchmarkMistral Medium 3.5Mistral Small 3.1 24BLeader
AA Intelligence Index14.2Not measuredOnly Mistral Medium 3.5 measured
Epoch Capabilities Index141.4Not measuredOnly Mistral Medium 3.5 measured
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index6 boards, none shared
Agentic work benchmarks for Mistral Medium 3.5 and Mistral Small 3.1 24B
BenchmarkMistral Medium 3.5Mistral Small 3.1 24BLeader
AA Agentic Index7.0Not measuredOnly Mistral Medium 3.5 measured
Finance Agent (Vals AI)32.1%Not measuredOnly Mistral Medium 3.5 measured
Tax Agent Bench (Vals AI)4.5%Not measuredOnly Mistral Medium 3.5 measured
Legal Research Bench (Vals AI)9.1%Not measuredOnly Mistral Medium 3.5 measured
Excel Modeling Benchmark (Vals AI)11.2%Not measuredOnly Mistral Medium 3.5 measured
τ²-Bench Telecom94.2%Not measuredOnly Mistral Medium 3.5 measured
Neither model has a result on the other 35 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index50
Reasoning benchmarks for Mistral Medium 3.5 and Mistral Small 3.1 24B
BenchmarkMistral Medium 3.5Mistral Small 3.1 24BLeader
GPQA Diamond34.9%Not measuredOnly Mistral Medium 3.5 measured
GPQA Diamond (Artificial Analysis run)74.8%Not measuredOnly Mistral Medium 3.5 measured
Humanity's Last Exam (Artificial Analysis run)13.8%Not measuredOnly Mistral Medium 3.5 measured
LMArena Text: hard prompts14361278Mistral Medium 3.5
LMArena Text: expert14321257Mistral Medium 3.5
LMArena Text: hard prompts in English14451290Mistral Medium 3.5
LMArena Vision: diagrams12541119Mistral Medium 3.5
LMArena Vision: homework12301158Mistral Medium 3.5
Kagi LLM Benchmark41.4%Not measuredOnly Mistral Medium 3.5 measured
Neither model has a result on the other 13 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index20
Math benchmarks for Mistral Medium 3.5 and Mistral Small 3.1 24B
BenchmarkMistral Medium 3.5Mistral Small 3.1 24BLeader
ProofBench9.0%Not measuredOnly Mistral Medium 3.5 measured
LMArena Text: math14311261Mistral Medium 3.5
LMArena Text: mathematical work14401271Mistral Medium 3.5
Neither model has a result on the other 10 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index40
Knowledge benchmarks for Mistral Medium 3.5 and Mistral Small 3.1 24B
BenchmarkMistral Medium 3.5Mistral Small 3.1 24BLeader
MMLU Pro75.3%Not measuredOnly Mistral Medium 3.5 measured
CaseLaw (Vals AI)44.2%Not measuredOnly Mistral Medium 3.5 measured
CorpFin (Vals AI)58.8%Not measuredOnly Mistral Medium 3.5 measured
TaxEval (Vals AI)68.0%Not measuredOnly Mistral Medium 3.5 measured
MortgageTax (Vals AI)28.9%Not measuredOnly Mistral Medium 3.5 measured
MedCode (Vals AI)33.8%Not measuredOnly Mistral Medium 3.5 measured
LMArena Text: science14281265Mistral Medium 3.5
LMArena Text: medicine and healthcare14091267Mistral Medium 3.5
LMArena Text: legal and government14371271Mistral Medium 3.5
LMArena Text: business and finance14261257Mistral Medium 3.5
Neither model has a result on the other 8 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
Web design benchmarks for Mistral Medium 3.5 and Mistral Small 3.1 24B
BenchmarkMistral Medium 3.5Mistral Small 3.1 24BLeader
Design Arena: websites1142Not measuredOnly Mistral Medium 3.5 measured
Design Arena: UI components1079Not measuredOnly Mistral Medium 3.5 measured
Design Arena: game dev1079Not measuredOnly Mistral Medium 3.5 measured
Design Arena: data viz1120Not measuredOnly Mistral Medium 3.5 measured
Design Arena: 3D1061Not measuredOnly Mistral Medium 3.5 measured
Design Arena: all code categories1121Not measuredOnly Mistral Medium 3.5 measured
Design Arena: SVG950Not measuredOnly Mistral Medium 3.5 measured
Design Arena: ASCII art1063Not measuredOnly Mistral Medium 3.5 measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index220
General benchmarks for Mistral Medium 3.5 and Mistral Small 3.1 24B
BenchmarkMistral Medium 3.5Mistral Small 3.1 24BLeader
SAGE (Vals AI)37.6%Not measuredOnly Mistral Medium 3.5 measured
MedScribe (Vals AI)67.7%Not measuredOnly Mistral Medium 3.5 measured
Vals Multimodal Index34.8%Not measuredOnly Mistral Medium 3.5 measured
IFBench68.8%Not measuredOnly Mistral Medium 3.5 measured
Long context reasoning (AA-LCR)69.3%Not measuredOnly Mistral Medium 3.5 measured
DTBench75.5%Not measuredOnly Mistral Medium 3.5 measured
LMCA26.1%Not measuredOnly Mistral Medium 3.5 measured
Surface Evolver Bench26.9%Not measuredOnly Mistral Medium 3.5 measured
LMArena Text14211277Mistral Medium 3.5
LMArena Text: instruction following14151263Mistral Medium 3.5
LMArena Text: creative writing13741252Mistral Medium 3.5
LMArena Vision12221136Mistral Medium 3.5
LMArena Text: English14331295Mistral Medium 3.5
LMArena Text: non-English14041255Mistral Medium 3.5
LMArena Text: Chinese14441253Mistral Medium 3.5
LMArena Text: German14531266Mistral Medium 3.5
LMArena Text: JapaneseNot measured1208Only Mistral Small 3.1 24B measured
LMArena Text: Korean13841207Mistral Medium 3.5
LMArena Text: French14481273Mistral Medium 3.5
LMArena Text: Spanish14111284Mistral Medium 3.5
LMArena Text: Russian13951263Mistral Medium 3.5
LMArena Text: Polish14301242Mistral Medium 3.5
LMArena Text: multi-turn14231270Mistral Medium 3.5
LMArena Text: longer queries14151299Mistral Medium 3.5
LMArena Text: writing and language13841268Mistral Medium 3.5
LMArena Text: entertainment and media13821250Mistral Medium 3.5
LMArena Vision: English12401157Mistral Medium 3.5
LMArena Vision: Chinese12441155Mistral Medium 3.5
LMArena Vision: reading text12331132Mistral Medium 3.5
LMArena Vision: humor12161142Mistral Medium 3.5
LMArena Vision: captioningNot measured1153Only Mistral Small 3.1 24B measured
LMArena Vision: recognizing thingsNot measured1155Only Mistral Small 3.1 24B measured
LMArena Vision: creative writing11981123Mistral Medium 3.5
Neither model has a result on the other 13 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Mistral Medium 3.5 and Mistral Small 3.1 24B
BenchmarkMistral Medium 3.5Mistral Small 3.1 24BLeader
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

Mistral Medium 3.5$0.0053Fits in one request
Mistral Small 3.1 24B$0.0006Fits in one request

Mistral Small 3.1 24B costs less

Repository review

50K input + 3K output tokens

Mistral Medium 3.5$0.098Fits in one request
Mistral Small 3.1 24B$0.019Fits in one request

Mistral Small 3.1 24B costs less

Agent loop

200K input + 10K output tokens

Mistral Medium 3.5$0.375Fits in one request
Mistral Small 3.1 24B$0.076Exceeds the context window

Mistral Small 3.1 24B 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 Mistral Medium 3.5 and Mistral Small 3.1 24B
SpecificationMistral Medium 3.5Mistral Small 3.1 24B
MakerMistral AIMistral AI
WeightsNo public weights listedOpen weights
ReleasedApril 30, 2026March 17, 2025
Context window262K128K
Input $/Mtok$1.50$0.351
Output $/Mtok$7.50$0.555
Coding index25.5Not benchmarked
Independent measurements behind it30
Cost per solved task$6.02Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is Mistral Medium 3.5 better than Mistral Small 3.1 24B for coding?

Only Mistral Medium 3.5 has a coding score, 25.5 on the Vibe Coding Index; Mistral Small 3.1 24B has none, so the two cannot be ranked on coding. Pick Mistral Small 3.1 24B for output tokens 13.5x cheaper ($0.555 vs $7.50 per million).

Which is cheaper, Mistral Medium 3.5 or Mistral Small 3.1 24B?

Mistral Medium 3.5 costs $1.50 per million input tokens and $7.50 per million output tokens; Mistral Small 3.1 24B costs $0.351 and $0.555. Cost per solved task is known for Mistral Medium 3.5 only.

Which has the bigger context window, Mistral Medium 3.5 or Mistral Small 3.1 24B?

Mistral Medium 3.5, with 262K tokens against 128K.

How many benchmarks is the Mistral Medium 3.5 vs Mistral Small 3.1 24B comparison based on?

35 benchmarks have a result for both models as of October 21, 2026. Mistral Medium 3.5's coding index rests on 3 independent benchmarks and Mistral Small 3.1 24B's on 0. Where only one model has a result, the page says so instead of naming a winner.

Compare next