GPT-5.2-Codex vs MiniMax M2.5
GPT-5.2-Codex leads MiniMax M2.5 on coding, 45.4 to 37.2 on the Vibe Coding Index, and MiniMax M2.5 costs 11.7x less per output token ($1.20 vs $14.00 per million). Updated October 22, 2026; 17 benchmarks have a result for both.
GPT-5.2-Codex or MiniMax M2.5: which to pick
Pick GPT-5.2-Codex for coding: it scores 45.4 on the Vibe Coding Index against 37.2 for MiniMax M2.5.
Pick MiniMax M2.5 for output tokens 11.7x cheaper ($1.20 vs $14.00 per million) and work measured by SWE-bench Verified or SWE-bench Multilingual, where it leads.
- Shared results
- 17
- GPT-5.2-Codex only
- 5
- MiniMax M2.5 only
- 51
- 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.
Coding
Vibe Coding Index, six public coding benchmarks
GPT-5.2-Codex
GPT-5.2-Codex scores 45.4 against 37.2.
3 benchmarks behind each score
Intelligence
AA Intelligence Index
GPT-5.2-Codex
GPT-5.2-Codex leads, 28.5 against 22.8.
Same board, both measured
Reasoning
GPQA Diamond
No call
Only MiniMax M2.5 has a GPQA Diamond result, 82.1%.
Directional only
Math
AIME
No call
Only MiniMax M2.5 has a AIME result, 88.8%.
Directional only
Knowledge
MMLU Pro
No call
Only MiniMax M2.5 has a MMLU Pro result, 80.1%.
Directional only
Web design
Design Arena: websites
No call
Only MiniMax M2.5 has a Design Arena: websites result, 1222.
Directional only
Long documents
Listed context window
GPT-5.2-Codex
GPT-5.2-Codex takes 400K tokens against 205K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
MiniMax M2.5
GPT-5.2-Codex $0.0088, MiniMax M2.5 $0.0009 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
GPT-5.2-Codex
GPT-5.2-Codex $0.43, MiniMax M2.5 $0.62.
Published attempt cost and pass rate
At a glance
- Coding index
- GPT-5.2-Codex 45.4 from 3 benchmarks, MiniMax M2.5 37.2 from 3 benchmarks.
- Widest category
- Coding, 10.2 points apart on average over 11 shared benchmarks.
- Widest single gap
- Terminal-Bench: GPT-5.2-Codex 66.5% against MiniMax M2.5 42.7%, 23.8 points apart on a 0 to 100 scale.
- Output price
- GPT-5.2-Codex $14.00 against MiniMax M2.5 $1.20 per million output tokens, 11.7x apart.
- Context window
- GPT-5.2-Codex takes 400K tokens, MiniMax M2.5 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.
| Category | GPT-5.2-Codex | Bars | MiniMax M2.5 | Ties |
|---|---|---|---|---|
| Coding | 5 | 6 | ||
| Intelligence | 1 | 0 | ||
| Agentic work | 0 | 1 | ||
| Reasoning | 2 | 0 | ||
| General | 2 | 0 | ||
| All shared boards | 10 | 7 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index56
| Benchmark | GPT-5.2-Codex | MiniMax M2.5 | Leader |
|---|---|---|---|
| SWE-bench Verified | 72.8% | 75.8% | MiniMax M2.5 |
| SWE-bench Multilingual | 66.3% | 68.3% | MiniMax M2.5 |
| LMArena WebDev | 1339 | 1386 | MiniMax M2.5 |
| LiveCodeBench (Vals AI) | 88.0% | 79.2% | GPT-5.2-Codex |
| SWE-bench Verified (Vals AI) | 72.4% | 74.2% | MiniMax M2.5 |
| Vibe Code Bench (Vals AI) | 37.9% | 14.8% | GPT-5.2-Codex |
| Terminal-Bench Hard (Artificial Analysis run) | 37.1% | 34.8% | GPT-5.2-Codex |
| Terminal-Bench | 66.5% | 42.7% | GPT-5.2-Codex |
| ALE-Bench | 1299.9 | 618.2 | GPT-5.2-Codex |
| LMArena Text: coding | Not measured | 1380 | Only MiniMax M2.5 measured |
| LMArena Text: software and IT | Not measured | 1383 | Only MiniMax M2.5 measured |
| LMArena WebDev: React | 1329 | 1374 | MiniMax M2.5 |
| LMArena WebDev: HTML | 1348 | 1414 | MiniMax M2.5 |
| Neither model has a result on the other 26 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | GPT-5.2-Codex | MiniMax M2.5 | Leader |
|---|---|---|---|
| AA Intelligence Index | 28.5 | 22.8 | GPT-5.2-Codex |
| Epoch Capabilities Index | Not measured | 146.6 | Only MiniMax M2.5 measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | GPT-5.2-Codex | MiniMax M2.5 | Leader |
|---|---|---|---|
| Design Arena: agents building web apps | 1051 | Not measured | Only GPT-5.2-Codex measured |
| Design Arena: agents building full-stack apps | 993 | Not measured | Only GPT-5.2-Codex measured |
| Design Arena: agents building mobile apps | 1090 | Not measured | Only GPT-5.2-Codex measured |
| Design Arena: agents building Android apps | 1176 | Not measured | Only GPT-5.2-Codex measured |
| Design Arena: agents building Godot games | 1142 | Not measured | Only GPT-5.2-Codex measured |
| τ²-Bench Telecom | 92.1% | 95.3% | MiniMax M2.5 |
| Vending-Bench 2 | Not measured | -$23 | Only MiniMax M2.5 measured |
| Neither model has a result on the other 34 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | GPT-5.2-Codex | MiniMax M2.5 | Leader |
|---|---|---|---|
| GPQA Diamond | Not measured | 82.1% | Only MiniMax M2.5 measured |
| GPQA Diamond (Artificial Analysis run) | 89.9% | 84.8% | GPT-5.2-Codex |
| Humanity's Last Exam (Artificial Analysis run) | 35.7% | 20.5% | GPT-5.2-Codex |
| ARC-AGI | Not measured | 63.7% | Only MiniMax M2.5 measured |
| ARC-AGI-2 | Not measured | 4.9% | Only MiniMax M2.5 measured |
| LMArena Text: hard prompts | Not measured | 1372 | Only MiniMax M2.5 measured |
| LMArena Text: expert | Not measured | 1378 | Only MiniMax M2.5 measured |
| LMArena Text: hard prompts in English | Not measured | 1384 | Only MiniMax M2.5 measured |
| Kagi LLM Benchmark | Not measured | 55.2% | Only MiniMax M2.5 measured |
| Neither model has a result on the other 13 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index4 boards, none shared
| Benchmark | GPT-5.2-Codex | MiniMax M2.5 | Leader |
|---|---|---|---|
| AIME | Not measured | 88.8% | Only MiniMax M2.5 measured |
| ProofBench | Not measured | 4.0% | Only MiniMax M2.5 measured |
| LMArena Text: math | Not measured | 1378 | Only MiniMax M2.5 measured |
| LMArena Text: mathematical work | Not measured | 1378 | Only MiniMax M2.5 measured |
| Neither model has a result on the other 9 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index10 boards, none shared
| Benchmark | GPT-5.2-Codex | MiniMax M2.5 | Leader |
|---|---|---|---|
| MMLU Pro | Not measured | 80.1% | Only MiniMax M2.5 measured |
| LegalBench (Vals AI) | Not measured | 80.0% | Only MiniMax M2.5 measured |
| CaseLaw (Vals AI) | Not measured | 53.5% | Only MiniMax M2.5 measured |
| CorpFin (Vals AI) | Not measured | 59.6% | Only MiniMax M2.5 measured |
| TaxEval (Vals AI) | Not measured | 68.2% | Only MiniMax M2.5 measured |
| MedQA (Vals AI) | Not measured | 92.5% | Only MiniMax M2.5 measured |
| LMArena Text: science | Not measured | 1369 | Only MiniMax M2.5 measured |
| LMArena Text: medicine and healthcare | Not measured | 1357 | Only MiniMax M2.5 measured |
| LMArena Text: legal and government | Not measured | 1366 | Only MiniMax M2.5 measured |
| LMArena Text: business and finance | Not measured | 1356 | Only MiniMax M2.5 measured |
| Neither model has a result on the other 8 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index7 boards, none shared
| Benchmark | GPT-5.2-Codex | MiniMax M2.5 | Leader |
|---|---|---|---|
| Design Arena: websites | Not measured | 1222 | Only MiniMax M2.5 measured |
| Design Arena: UI components | Not measured | 1173 | Only MiniMax M2.5 measured |
| Design Arena: game dev | Not measured | 1176 | Only MiniMax M2.5 measured |
| Design Arena: data viz | Not measured | 1172 | Only MiniMax M2.5 measured |
| Design Arena: 3D | Not measured | 1170 | Only MiniMax M2.5 measured |
| Design Arena: all code categories | Not measured | 1207 | Only MiniMax M2.5 measured |
| Design Arena: SVG | Not measured | 1150 | Only MiniMax M2.5 measured |
| Neither model has a result on the other 2 boards here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | GPT-5.2-Codex | MiniMax M2.5 | Leader |
|---|---|---|---|
| IFBench | 77.6% | 71.6% | GPT-5.2-Codex |
| Long context reasoning (AA-LCR) | 82.3% | 73.3% | GPT-5.2-Codex |
| CL-bench | Not measured | 11.4% | Only MiniMax M2.5 measured |
| CL-bench Life | Not measured | 6.3% | Only MiniMax M2.5 measured |
| LMArena Text | Not measured | 1359 | Only MiniMax M2.5 measured |
| LMArena Text: instruction following | Not measured | 1353 | Only MiniMax M2.5 measured |
| LMArena Text: creative writing | Not measured | 1331 | Only MiniMax M2.5 measured |
| LMArena Text: English | Not measured | 1380 | Only MiniMax M2.5 measured |
| LMArena Text: non-English | Not measured | 1338 | Only MiniMax M2.5 measured |
| LMArena Text: Chinese | Not measured | 1393 | Only MiniMax M2.5 measured |
| LMArena Text: German | Not measured | 1361 | Only MiniMax M2.5 measured |
| LMArena Text: Japanese | Not measured | 1171 | Only MiniMax M2.5 measured |
| LMArena Text: Korean | Not measured | 1233 | Only MiniMax M2.5 measured |
| LMArena Text: French | Not measured | 1363 | Only MiniMax M2.5 measured |
| LMArena Text: Spanish | Not measured | 1353 | Only MiniMax M2.5 measured |
| LMArena Text: Russian | Not measured | 1359 | Only MiniMax M2.5 measured |
| LMArena Text: Polish | Not measured | 1334 | Only MiniMax M2.5 measured |
| LMArena Text: multi-turn | Not measured | 1364 | Only MiniMax M2.5 measured |
| LMArena Text: longer queries | Not measured | 1366 | Only MiniMax M2.5 measured |
| LMArena Text: writing and language | Not measured | 1345 | Only MiniMax M2.5 measured |
| LMArena Text: entertainment and media | Not measured | 1329 | Only MiniMax M2.5 measured |
| Neither model has a result on the other 25 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | GPT-5.2-Codex | MiniMax M2.5 | 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
MiniMax M2.5 costs less
Repository review
50K input + 3K output tokens
MiniMax M2.5 costs less
Agent loop
200K input + 10K output tokens
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.
| Specification | OpenAI GPT-5.2-Codex | MiniMax M2.5 |
|---|---|---|
| Maker | OpenAI | MiniMax |
| Weights | No public weights listed | Open weights |
| Released | January 14, 2026 | February 12, 2026 |
| Context window | 400K | 205K |
| Input $/Mtok | $1.75 | $0.30 |
| Output $/Mtok | $14.00 | $1.20 |
| Coding index | 45.4 | 37.2 |
| Independent measurements behind it | 3 | 3 |
| Cost per solved task | $0.43 | $0.62 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is GPT-5.2-Codex better than MiniMax M2.5 for coding?
Pick GPT-5.2-Codex for coding: it scores 45.4 on the Vibe Coding Index against 37.2 for MiniMax M2.5. Pick MiniMax M2.5 for output tokens 11.7x cheaper ($1.20 vs $14.00 per million) and work measured by SWE-bench Verified or SWE-bench Multilingual, where it leads.
Which is cheaper, GPT-5.2-Codex or MiniMax M2.5?
GPT-5.2-Codex costs $1.75 per million input tokens and $14.00 per million output tokens; MiniMax M2.5 costs $0.30 and $1.20. Per solved SWE-bench Verified task, GPT-5.2-Codex costs $0.43 and MiniMax M2.5 $0.62.
Which has the bigger context window, GPT-5.2-Codex or MiniMax M2.5?
GPT-5.2-Codex, with 400K tokens against 205K.
How many benchmarks is the GPT-5.2-Codex vs MiniMax M2.5 comparison based on?
17 benchmarks have a result for both models as of October 22, 2026. GPT-5.2-Codex's coding index rests on 3 independent benchmarks and MiniMax M2.5's on 3. Where only one model has a result, the page says so instead of naming a winner.