Enter opens the highlighted result. Esc closes.

Phi 4 vs Mistral Medium 3.1

Mistral Medium 3.1 leads Phi 4 on the Artificial Analysis Intelligence Index, 9.2 to 5.9, and Phi 4 costs 14.3x less per output token ($0.14 vs $2.00 per million). Updated October 21, 2026; 38 benchmarks have a result for both.

Replace Phi 4 Replace Mistral Medium 3.1
Phi 4Microsoft
Coding
No data
Overall
24.7
Output $/M
$0.14
1
Mistral Medium 3.1Mistral AI
Coding
No data
Overall
No data
Output $/M
$2.00
37
vs boards won of 38 shared

Phi 4 or Mistral Medium 3.1: which to pick

Neither model has a coding score yet, so the two cannot be ranked on coding.

Pick Phi 4 for output tokens 14.3x cheaper ($0.14 vs $2.00 per million) and work measured by LiveCodeBench (Artificial Analysis run) or Terminal-Bench Hard (Artificial Analysis run), where it leads.

Model APhi 4MicrosoftNot benchmarkedNo coding benchmarkNo cost per solved taskModel BMistral Medium 3.1Mistral AINot benchmarkedNo coding benchmarkNo cost per solved task
Shared results
38
Phi 4 only
8
Mistral Medium 3.1 only
36
Like-for-like categories
7 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
Phi 4 not benchmarked, Mistral Medium 3.1 not benchmarked.
Widest category
Agentic work, 40.6 points apart on average over 1 shared benchmark.
Widest single gap
Long context reasoning (AA-LCR): Phi 4 0.0% against Mistral Medium 3.1 42.7%, 42.7 points apart on a 0 to 100 scale.
Output price
Phi 4 $0.14 against Mistral Medium 3.1 $2.00 per million output tokens, 14.3x apart.
Context window
Phi 4 takes 16K tokens, Mistral Medium 3.1 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.

Boards won, where both have a result
CategoryPhi 4BarsMistral Medium 3.1Ties
Coding04
Intelligence01
Agentic work01
Reasoning05
Math04
Knowledge14
General018
All shared boards137
CodingLike-for-likeSix of these boards feed the Vibe Coding Index04
Coding benchmarks for Phi 4 and Mistral Medium 3.1
BenchmarkPhi 4Mistral Medium 3.1Leader
AA Coding IndexNot measured20.5Only Mistral Medium 3.1 measured
LiveCodeBench (Vals AI)Not measured44.8%Only Mistral Medium 3.1 measured
LiveCodeBench (Artificial Analysis run)23.1%40.6%Mistral Medium 3.1
SciCodeNot measured32.4%Only Mistral Medium 3.1 measured
Terminal-Bench Hard (Artificial Analysis run)3.8%10.6%Mistral Medium 3.1
ALE-BenchNot measured210.2Only Mistral Medium 3.1 measured
LMArena Text: coding12311433Mistral Medium 3.1
LMArena Text: software and IT12281436Mistral Medium 3.1
Neither model has a result on the other 31 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01
Intelligence benchmarks for Phi 4 and Mistral Medium 3.1
BenchmarkPhi 4Mistral Medium 3.1Leader
AA Intelligence Index5.99.2Mistral Medium 3.1
Epoch Capabilities Index130.5Not measuredOnly Phi 4 measured
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
Agentic work benchmarks for Phi 4 and Mistral Medium 3.1
BenchmarkPhi 4Mistral Medium 3.1Leader
AA Agentic IndexNot measured1.9Only Mistral Medium 3.1 measured
τ²-Bench Telecom0.0%40.6%Mistral Medium 3.1
BALROG11.6%Not measuredOnly Phi 4 measured
Berkeley Function Calling Leaderboard (BFCL)28.8%Not measuredOnly Phi 4 measured
Neither model has a result on the other 37 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index05
Reasoning benchmarks for Phi 4 and Mistral Medium 3.1
BenchmarkPhi 4Mistral Medium 3.1Leader
MMMU Pro (Vals AI)Not measured63.0%Only Mistral Medium 3.1 measured
GPQA Diamond (Artificial Analysis run)57.5%58.8%Mistral Medium 3.1
Humanity's Last Exam (Artificial Analysis run)3.8%4.7%Mistral Medium 3.1
GPQA Diamond (Epoch AI run)56.1%Not measuredOnly Phi 4 measured
LMArena Text: hard prompts12201426Mistral Medium 3.1
LMArena Text: expert12031408Mistral Medium 3.1
LMArena Text: hard prompts in English12301435Mistral Medium 3.1
LMArena Vision: diagramsNot measured1175Only Mistral Medium 3.1 measured
LMArena Vision: homeworkNot measured1165Only Mistral Medium 3.1 measured
Neither model has a result on the other 13 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index04
Math benchmarks for Phi 4 and Mistral Medium 3.1
BenchmarkPhi 4Mistral Medium 3.1Leader
AIMENot measured42.3%Only Mistral Medium 3.1 measured
MATH 500 (Vals AI)Not measured87.0%Only Mistral Medium 3.1 measured
MGSM (Vals AI)Not measured90.9%Only Mistral Medium 3.1 measured
AA Math Index18.038.3Mistral Medium 3.1
AIME 2025 (Artificial Analysis run)18.0%38.3%Mistral Medium 3.1
OTIS Mock AIME13.8%Not measuredOnly Phi 4 measured
MATH level 564.9%Not measuredOnly Phi 4 measured
LMArena Text: math12461408Mistral Medium 3.1
LMArena Text: mathematical work12541412Mistral Medium 3.1
Neither model has a result on the other 4 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index14
Knowledge benchmarks for Phi 4 and Mistral Medium 3.1
BenchmarkPhi 4Mistral Medium 3.1Leader
MMLU ProNot measured75.3%Only Mistral Medium 3.1 measured
LegalBench (Vals AI)Not measured62.0%Only Mistral Medium 3.1 measured
CorpFin (Vals AI)Not measured50.7%Only Mistral Medium 3.1 measured
TaxEval (Vals AI)Not measured70.3%Only Mistral Medium 3.1 measured
MortgageTax (Vals AI)Not measured36.5%Only Mistral Medium 3.1 measured
MedQA (Vals AI)Not measured78.2%Only Mistral Medium 3.1 measured
MMLU Pro (Artificial Analysis run)71.4%68.3%Phi 4
MMLU84.8%Not measuredOnly Phi 4 measured
LMArena Text: science12061444Mistral Medium 3.1
LMArena Text: medicine and healthcare12021449Mistral Medium 3.1
LMArena Text: legal and government12411439Mistral Medium 3.1
LMArena Text: business and finance11991418Mistral Medium 3.1
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
Web design benchmarks for Phi 4 and Mistral Medium 3.1
BenchmarkPhi 4Mistral Medium 3.1Leader
Design Arena: websitesNot measured1133Only Mistral Medium 3.1 measured
Design Arena: UI componentsNot measured1100Only Mistral Medium 3.1 measured
Design Arena: game devNot measured1074Only Mistral Medium 3.1 measured
Design Arena: data vizNot measured1144Only Mistral Medium 3.1 measured
Design Arena: 3DNot measured1081Only Mistral Medium 3.1 measured
Design Arena: all code categoriesNot measured1120Only Mistral Medium 3.1 measured
Design Arena: SVGNot measured993Only Mistral Medium 3.1 measured
Design Arena: ASCII artNot measured997Only Mistral Medium 3.1 measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index018
General benchmarks for Phi 4 and Mistral Medium 3.1
BenchmarkPhi 4Mistral Medium 3.1Leader
IFBench23.5%39.8%Mistral Medium 3.1
Long context reasoning (AA-LCR)0.0%42.7%Mistral Medium 3.1
DTBenchNot measured67.5%Only Mistral Medium 3.1 measured
LMCANot measured19.1%Only Mistral Medium 3.1 measured
Creative writing (Lech Mazur)62.6%Not measuredOnly Phi 4 measured
LMArena Text12171424Mistral Medium 3.1
LMArena Text: instruction following12011398Mistral Medium 3.1
LMArena Text: creative writing11811391Mistral Medium 3.1
LMArena VisionNot measured1172Only Mistral Medium 3.1 measured
LMArena Text: English12311438Mistral Medium 3.1
LMArena Text: non-English11971408Mistral Medium 3.1
LMArena Text: Chinese12121447Mistral Medium 3.1
LMArena Text: German12221432Mistral Medium 3.1
LMArena Text: Japanese11581379Mistral Medium 3.1
LMArena Text: Korean11511380Mistral Medium 3.1
LMArena Text: French12241460Mistral Medium 3.1
LMArena Text: Spanish12341433Mistral Medium 3.1
LMArena Text: Russian12091411Mistral Medium 3.1
LMArena Text: PolishNot measured1430Only Mistral Medium 3.1 measured
LMArena Text: multi-turn12061418Mistral Medium 3.1
LMArena Text: longer queries12171406Mistral Medium 3.1
LMArena Text: writing and language11981395Mistral Medium 3.1
LMArena Text: entertainment and media11621390Mistral Medium 3.1
LMArena Vision: EnglishNot measured1179Only Mistral Medium 3.1 measured
LMArena Vision: ChineseNot measured1190Only Mistral Medium 3.1 measured
LMArena Vision: reading textNot measured1175Only Mistral Medium 3.1 measured
LMArena Vision: humorNot measured1168Only Mistral Medium 3.1 measured
LMArena Vision: captioningNot measured1145Only Mistral Medium 3.1 measured
LMArena Vision: recognizing thingsNot measured1147Only Mistral Medium 3.1 measured
LMArena Vision: creative writingNot measured1200Only Mistral Medium 3.1 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
Image generation benchmarks for Phi 4 and Mistral Medium 3.1
BenchmarkPhi 4Mistral Medium 3.1Leader
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

Phi 4$0.0001Fits in one request
Mistral Medium 3.1$0.0014Fits in one request

Phi 4 costs less

Repository review

50K input + 3K output tokens

Phi 4$0.0039Exceeds the context window
Mistral Medium 3.1$0.026Fits in one request

Phi 4 costs less

Agent loop

200K input + 10K output tokens

Phi 4$0.015Exceeds the context window
Mistral Medium 3.1$0.100Exceeds the context window

Phi 4 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 Phi 4 and Mistral Medium 3.1
SpecificationMicrosoft Phi 4Mistral Medium 3.1
MakerMicrosoftMistral AI
WeightsOpen weightsNo public weights listed
ReleasedJanuary 10, 2025August 13, 2025
Context window16K131K
Input $/Mtok$0.07$0.40
Output $/Mtok$0.14$2.00
Coding indexNot benchmarkedNot benchmarked
Independent measurements behind it00
Cost per solved taskNot computableNot computable
SourceOpenRouter listingOpenRouter listing

Questions

Is Phi 4 better than Mistral Medium 3.1 for coding?

Neither model has a coding score yet, so the two cannot be ranked on coding. Pick Phi 4 for output tokens 14.3x cheaper ($0.14 vs $2.00 per million) and work measured by LiveCodeBench (Artificial Analysis run) or Terminal-Bench Hard (Artificial Analysis run), where it leads.

Which is cheaper, Phi 4 or Mistral Medium 3.1?

Phi 4 costs $0.07 per million input tokens and $0.14 per million output tokens; Mistral Medium 3.1 costs $0.40 and $2.00. Cost per solved task is known for neither of them.

Which has the bigger context window, Phi 4 or Mistral Medium 3.1?

Mistral Medium 3.1, with 131K tokens against 16K.

How many benchmarks is the Phi 4 vs Mistral Medium 3.1 comparison based on?

38 benchmarks have a result for both models as of October 21, 2026. Phi 4's coding index rests on 0 independent benchmarks and Mistral Medium 3.1's on 0. Where only one model has a result, the page says so instead of naming a winner.

Compare next