LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

GPT-4o vs MiniMax M2

Updated October 21, 2026. 17 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace GPT-4o Replace MiniMax M2
GPT-4oOpenAI
Coding
38.8
Overall
No data
Output $/M
$10.00
0
MiniMax M2MiniMax
Coding
45.3
Overall
18.6
Output $/M
$1.02
17
vs boards won of 17 shared

Decision reading

Pick MiniMax M2 for coding: it scores 45.3 on the Vibe Coding Index against 38.8 for GPT-4o.

GPT-4o is neither cheaper nor larger in context, and leads no benchmark both models were measured on.

Model AGPT-4oOpenAI38.8/1001 independent benchmarkNo cost per solved taskModel BMiniMax M2MiniMax45.3/1002 independent benchmarks$0.702 per solved task
Shared results
17
GPT-4o only
9
MiniMax M2 only
32
Like-for-like categories
7 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.

At a glance

Coding index
GPT-4o 38.8 from 1 benchmark, MiniMax M2 45.3 from 2 benchmarks.
Widest category
Math, 72.3 points apart on average over 2 shared benchmarks.
Widest single gap
AIME 2025 (Artificial Analysis run): GPT-4o 6.0% against MiniMax M2 78.3%, 72.3 points apart on a 0 to 100 scale.
Output price
GPT-4o $10.00 against MiniMax M2 $1.02 per million output tokens, 9.8x apart.
Context window
GPT-4o takes 128K tokens, MiniMax M2 205K.

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
CategoryGPT-4oBarsMiniMax M2Ties
Coding03
Agentic work01
Reasoning02
Math02
Knowledge01
Web design06
General02
All shared boards017
CodingLike-for-likeFour of these boards feed the Vibe Coding Index03
Coding benchmarks for GPT-4o and MiniMax M2
BenchmarkGPT-4oMiniMax M2Leader
SWE-bench Verified38.8%61.0%MiniMax M2
SWE-bench Lite38.0%Not measuredOnly GPT-4o measured
SWE-bench (full test set)18.8%Not measuredOnly GPT-4o measured
SWE-bench Multimodal30.4%Not measuredOnly GPT-4o measured
LMArena WebDevNot measured1297Only MiniMax M2 measured
LiveCodeBench (Artificial Analysis run)30.9%82.6%MiniMax M2
Terminal-Bench Hard (Artificial Analysis run)8.3%25.8%MiniMax M2
Terminal-BenchNot measured30.0%Only MiniMax M2 measured
LMArena Text: codingNot measured1371Only MiniMax M2 measured
LMArena Text: software and ITNot measured1373Only MiniMax M2 measured
LMArena WebDev: HTMLNot measured1306Only MiniMax M2 measured
Neither model has a result on the other 27 boards here
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index1 board, none shared
Intelligence benchmarks for GPT-4o and MiniMax M2
BenchmarkGPT-4oMiniMax M2Leader
AA Intelligence IndexNot measured18.6Only MiniMax M2 measured
Neither model has a result on the other 1 board here
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
Agentic work benchmarks for GPT-4o and MiniMax M2
BenchmarkGPT-4oMiniMax M2Leader
τ²-Bench Telecom25.1%86.8%MiniMax M2
GDPval9.9%Not measuredOnly GPT-4o measured
Vending-Bench 2Not measured$161Only MiniMax M2 measured
Neither model has a result on the other 36 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index02
Reasoning benchmarks for GPT-4o and MiniMax M2
BenchmarkGPT-4oMiniMax M2Leader
GPQA Diamond (Artificial Analysis run)54.3%77.7%MiniMax M2
Humanity's Last Exam (Artificial Analysis run)2.4%13.7%MiniMax M2
ARC-AGI4.5%Not measuredOnly GPT-4o measured
ARC-AGI-20.0%Not measuredOnly GPT-4o measured
LMArena Text: hard promptsNot measured1357Only MiniMax M2 measured
LMArena Text: expertNot measured1338Only MiniMax M2 measured
LMArena Text: hard prompts in EnglishNot measured1376Only MiniMax M2 measured
Neither model has a result on the other 14 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index02
Math benchmarks for GPT-4o and MiniMax M2
BenchmarkGPT-4oMiniMax M2Leader
AA Math Index6.078.3MiniMax M2
AIME 2025 (Artificial Analysis run)6.0%78.3%MiniMax M2
LMArena Text: mathNot measured1351Only MiniMax M2 measured
LMArena Text: mathematical workNot measured1368Only MiniMax M2 measured
Neither model has a result on the other 9 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index01
Knowledge benchmarks for GPT-4o and MiniMax M2
BenchmarkGPT-4oMiniMax M2Leader
MMLU Pro (Artificial Analysis run)74.8%82.0%MiniMax M2
MMLU88.1%Not measuredOnly GPT-4o measured
LMArena Text: scienceNot measured1348Only MiniMax M2 measured
LMArena Text: medicine and healthcareNot measured1367Only MiniMax M2 measured
LMArena Text: legal and governmentNot measured1349Only MiniMax M2 measured
LMArena Text: business and financeNot measured1336Only MiniMax M2 measured
Neither model has a result on the other 11 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index06
Web design benchmarks for GPT-4o and MiniMax M2
BenchmarkGPT-4oMiniMax M2Leader
Design Arena: websites8311144MiniMax M2
Design Arena: UI components8851136MiniMax M2
Design Arena: game dev9061119MiniMax M2
Design Arena: data viz8581140MiniMax M2
Design Arena: 3D8691098MiniMax M2
Design Arena: all code categories8611135MiniMax M2
Design Arena: SVGNot measured1098Only MiniMax M2 measured
Neither model has a result on the other 2 boards here
GeneralLike-for-likePublic boards, shown as published, kept out of the index02
General benchmarks for GPT-4o and MiniMax M2
BenchmarkGPT-4oMiniMax M2Leader
IFBench34.3%72.3%MiniMax M2
Long context reasoning (AA-LCR)49.3%64.3%MiniMax M2
DTBench64.5%Not measuredOnly GPT-4o measured
LMCA16.6%Not measuredOnly GPT-4o measured
LMArena TextNot measured1340Only MiniMax M2 measured
LMArena Text: instruction followingNot measured1329Only MiniMax M2 measured
LMArena Text: creative writingNot measured1286Only MiniMax M2 measured
LMArena Text: EnglishNot measured1364Only MiniMax M2 measured
LMArena Text: non-EnglishNot measured1313Only MiniMax M2 measured
LMArena Text: ChineseNot measured1366Only MiniMax M2 measured
LMArena Text: GermanNot measured1357Only MiniMax M2 measured
LMArena Text: FrenchNot measured1336Only MiniMax M2 measured
LMArena Text: SpanishNot measured1326Only MiniMax M2 measured
LMArena Text: RussianNot measured1330Only MiniMax M2 measured
LMArena Text: PolishNot measured1312Only MiniMax M2 measured
LMArena Text: multi-turnNot measured1361Only MiniMax M2 measured
LMArena Text: longer queriesNot measured1331Only MiniMax M2 measured
LMArena Text: writing and languageNot measured1295Only MiniMax M2 measured
LMArena Text: entertainment and mediaNot measured1298Only MiniMax M2 measured
Neither model has a result on the other 27 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for GPT-4o and MiniMax M2
BenchmarkGPT-4oMiniMax M2Leader
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

GPT-4o$0.0075Fits in one request
MiniMax M2$0.0008Fits in one request

MiniMax M2 costs less

Repository review

50K input + 3K output tokens

GPT-4o$0.155Fits in one request
MiniMax M2$0.016Fits in one request

MiniMax M2 costs less

Agent loop

200K input + 10K output tokens

GPT-4o$0.600Exceeds the context window
MiniMax M2$0.061Exceeds the context window

MiniMax M2 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 GPT-4o and MiniMax M2
SpecificationOpenAI GPT-4oMiniMax M2
MakerOpenAIMiniMax
WeightsNo public weights listedOpen weights
ReleasedMay 13, 2024October 23, 2025
Context window128K205K
Input $/Mtok$2.50$0.255
Output $/Mtok$10.00$1.02
Coding index38.845.3
Independent measurements behind it12
Cost per solved taskNot computable$0.702
SourceOpenRouter listingOpenRouter listing

Questions

Is GPT-4o better than MiniMax M2 for coding?

Pick MiniMax M2 for coding: it scores 45.3 on the Vibe Coding Index against 38.8 for GPT-4o. GPT-4o is neither cheaper nor larger in context, and leads no benchmark both models were measured on.

Which is cheaper, GPT-4o or MiniMax M2?

GPT-4o costs $2.50 per million input tokens and $10.00 per million output tokens; MiniMax M2 costs $0.255 and $1.02. Cost per solved task is known for MiniMax M2 only.

Which has the bigger context window, GPT-4o or MiniMax M2?

MiniMax M2, with 205K tokens against 128K.

How many benchmarks is the GPT-4o vs MiniMax M2 comparison based on?

17 benchmarks have a result for both models as of October 21, 2026. GPT-4o's coding index rests on 1 independent benchmark and MiniMax M2's on 2. Where only one model has a result, the page says so instead of naming a winner.

Compare next