GPT-5.2 vs GPT-6.1 Sol
Updated October 21, 2026. 62 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.
vs boards won of 62 shared
Decision reading
Pick GPT-6.1 Sol for coding: it scores 88.7 on the Vibe Coding Index against 58.9 for GPT-5.2.
Pick GPT-5.2 when what matters is MedCode (Vals AI) or SAGE (Vals AI), where it leads.
Model AGPT-5.2OpenAI58.9/1002 independent benchmarks$1.91 per solved taskModel BGPT-6.1 SolOpenAI88.7/1002 independent benchmarksNo cost per solved task
- Shared results
- 62
- GPT-5.2 only
- 49
- GPT-6.1 Sol only
- 33
- Like-for-like categories
- 8 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.
Coding
Vibe Coding Index, four public coding benchmarks
GPT-6.1 Sol
GPT-6.1 Sol scores 88.7 against 58.9.
2 benchmarks behind each score
Intelligence
AA Intelligence Index
GPT-6.1 Sol
GPT-6.1 Sol leads, 51.8 against 30.4.
Same board, both measured
Reasoning
GPQA Diamond
No call
Only GPT-5.2 has a GPQA Diamond result, 91.7%.
Directional only
Math
AIME
No call
Only GPT-5.2 has a AIME result, 96.9%.
Directional only
Knowledge
MMLU Pro
No call
Only GPT-5.2 has a MMLU Pro result, 86.2%.
Directional only
Web design
Design Arena: websites
GPT-6.1 Sol
GPT-6.1 Sol leads, 1321 against 1196.
Same board, both measured
General
Vals Index
No call
Only GPT-6.1 Sol has a Vals Index result, 63.9%.
Directional only
Long documents
Listed context window
GPT-6.1 Sol
GPT-6.1 Sol takes 1.05M tokens against 400K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
GPT-6.1 Sol
GPT-5.2 $0.0088, GPT-6.1 Sol $0.0070 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only GPT-5.2 has the number, $1.91.
Directional only
At a glance
- Coding index
- GPT-5.2 58.9 from 2 benchmarks, GPT-6.1 Sol 88.7 from 2 benchmarks.
- Widest category
- Math, 33.8 points apart on average over 6 shared benchmarks.
- Widest single gap
- ProofBench: GPT-5.2 15.0% against GPT-6.1 Sol 99.0%, 84.0 points apart on a 0 to 100 scale.
- Output price
- GPT-5.2 $14.00 against GPT-6.1 Sol $10.00 per million output tokens, 1.4x apart.
- Context window
- GPT-5.2 takes 400K tokens, GPT-6.1 Sol 1.05M.
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 | Bars | GPT-6.1 Sol | Ties |
|---|---|---|---|---|
| Coding | 0 | 6 | ||
| Intelligence | 0 | 2 | ||
| Agentic work | 0 | 1 | ||
| Reasoning | 0 | 12 | ||
| Math | 0 | 6 | ||
| Knowledge | 1 | 5 | ||
| Web design | 0 | 6 | ||
| General | 1 | 22 | ||
| All shared boards | 2 | 60 |
CodingLike-for-likeFour of these boards feed the Vibe Coding Index06
| Benchmark | GPT-5.2 | GPT-6.1 Sol | Leader |
|---|---|---|---|
| SWE-bench Verified | 72.8% | Not measured | Only GPT-5.2 measured |
| SWE-bench Multilingual | 66.7% | Not measured | Only GPT-5.2 measured |
| LiveBench coding | Not measured | 96.0% | Only GPT-6.1 Sol measured |
| LMArena WebDev | 1417 | 1753 | GPT-6.1 Sol |
| LiveCodeBench (Vals AI) | 85.4% | Not measured | Only GPT-5.2 measured |
| SWE-bench Verified (Vals AI) | 75.8% | Not measured | Only GPT-5.2 measured |
| Terminal-Bench 4 (Vals AI) | Not measured | 55.0% | Only GPT-6.1 Sol measured |
| Vibe Code Bench (Vals AI) | 53.5% | 88.9% | GPT-6.1 Sol |
| Code migration (Vals AI) | Not measured | 65.1% | Only GPT-6.1 Sol measured |
| ProgramBench (Vals AI) | Not measured | 3.0% | Only GPT-6.1 Sol measured |
| IOI (Vals AI) | Not measured | 96.9% | Only GPT-6.1 Sol measured |
| SWE-bench Multimodal (Vals AI) | Not measured | 43.5% | Only GPT-6.1 Sol measured |
| CyberBench (Vals AI) | Not measured | 39.3% | Only GPT-6.1 Sol measured |
| LiveCodeBench (Artificial Analysis run) | 89.4% | Not measured | Only GPT-5.2 measured |
| SciCode | Not measured | 55.8% | Only GPT-6.1 Sol measured |
| Terminal-Bench Hard (Artificial Analysis run) | 47.0% | Not measured | Only GPT-5.2 measured |
| Terminal-Bench | 64.9% | Not measured | Only GPT-5.2 measured |
| FrontierCode | Not measured | 50.2% | Only GPT-6.1 Sol measured |
| GSO-Bench | 27.5% | Not measured | Only GPT-5.2 measured |
| WeirdML | 72.2% | Not measured | Only GPT-5.2 measured |
| ALE-Bench | 1293.5 | 2743.8 | GPT-6.1 Sol |
| LMArena Text: coding | 1442 | 1478 | GPT-6.1 Sol |
| LMArena Text: software and IT | 1434 | 1476 | GPT-6.1 Sol |
| LMArena WebDev: React | Not measured | 1754 | Only GPT-6.1 Sol measured |
| LMArena WebDev: HTML | 1427 | 1800 | GPT-6.1 Sol |
| LMArena WebDev: from an image | Not measured | 1702 | Only GPT-6.1 Sol measured |
| Neither model has a result on the other 12 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | GPT-5.2 | GPT-6.1 Sol | Leader |
|---|---|---|---|
| AA Intelligence Index | 30.4 | 51.8 | GPT-6.1 Sol |
| Epoch Capabilities Index | 153.7 | 168.4 | GPT-6.1 Sol |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | GPT-5.2 | GPT-6.1 Sol | Leader |
|---|---|---|---|
| Design Arena: agents building web apps | 1086 | Not measured | Only GPT-5.2 measured |
| Design Arena: agents building full-stack apps | 1043 | Not measured | Only GPT-5.2 measured |
| Design Arena: agents building mobile apps | 1094 | 1190 | GPT-6.1 Sol |
| Design Arena: agents building Android apps | 1064 | Not measured | Only GPT-5.2 measured |
| Design Arena: agents building Godot games | 1142 | Not measured | Only GPT-5.2 measured |
| Design Arena: spreadsheets | Not measured | 1250 | Only GPT-6.1 Sol measured |
| LiveBench research | Not measured | 83.5% | Only GPT-6.1 Sol measured |
| LiveBench browser | Not measured | 88.0% | Only GPT-6.1 Sol measured |
| LiveBench long agentic | Not measured | 73.0% | Only GPT-6.1 Sol measured |
| SREBench (Vals AI) | Not measured | 50.8% | Only GPT-6.1 Sol measured |
| Finance Agent (Vals AI) | Not measured | 52.0% | Only GPT-6.1 Sol measured |
| Tax Agent Bench (Vals AI) | Not measured | 21.8% | Only GPT-6.1 Sol measured |
| Legal Research Bench (Vals AI) | Not measured | 38.5% | Only GPT-6.1 Sol measured |
| Harvey's Legal Agent Benchmark (Vals AI) | Not measured | 6.3% | Only GPT-6.1 Sol measured |
| Excel Modeling Benchmark (Vals AI) | Not measured | 70.8% | Only GPT-6.1 Sol measured |
| BioMysteryBench (Vals AI) | Not measured | 79.6% | Only GPT-6.1 Sol measured |
| Terminal-Bench-Science (Vals AI) | Not measured | 52.9% | Only GPT-6.1 Sol measured |
| Vals Science Index | Not measured | 59.6% | Only GPT-6.1 Sol measured |
| Robo-bench (Vals AI) | Not measured | 28.3% | Only GPT-6.1 Sol measured |
| τ²-Bench Telecom | 84.8% | Not measured | Only GPT-5.2 measured |
| APEX-Agents | Not measured | 60.0% | Only GPT-6.1 Sol measured |
| METR time horizons | 75.3% | Not measured | Only GPT-5.2 measured |
| Remote Labor Index | 2.5% | Not measured | Only GPT-5.2 measured |
| GDPval | 49.7% | Not measured | Only GPT-5.2 measured |
| DeepResearch Bench | 41.1% | Not measured | Only GPT-5.2 measured |
| SpireClimb | Not measured | 24.7% | Only GPT-6.1 Sol measured |
| Vending-Bench 2 | $3,591 | Not measured | Only GPT-5.2 measured |
| Neither model has a result on the other 12 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index012
| Benchmark | GPT-5.2 | GPT-6.1 Sol | Leader |
|---|---|---|---|
| GPQA Diamond | 91.7% | Not measured | Only GPT-5.2 measured |
| LiveBench reasoning | Not measured | 92.6% | Only GPT-6.1 Sol measured |
| MysteryMechanism (Vals AI) | Not measured | 46.4% | Only GPT-6.1 Sol measured |
| MMMU Pro (Vals AI) | 86.7% | Not measured | Only GPT-5.2 measured |
| GPQA Diamond (Artificial Analysis run) | 90.3% | Not measured | Only GPT-5.2 measured |
| Humanity's Last Exam (Artificial Analysis run) | 37.7% | 52.9% | GPT-6.1 Sol |
| GPQA Diamond (Epoch AI run) | 91.4% | 95.4% | GPT-6.1 Sol |
| ARC-AGI | 86.2% | 98.5% | GPT-6.1 Sol |
| ARC-AGI-2 | 52.9% | 94.2% | GPT-6.1 Sol |
| SimpleBench | 45.8% | Not measured | Only GPT-5.2 measured |
| Chess puzzles | 48.0% | 68.0% | GPT-6.1 Sol |
| Mystery game puzzles | 23.0% | 80.0% | GPT-6.1 Sol |
| EBR-bench | 23.0% | 54.3% | GPT-6.1 Sol |
| VPCT | 84.0% | Not measured | Only GPT-5.2 measured |
| LMArena Text: hard prompts | 1428 | 1463 | GPT-6.1 Sol |
| LMArena Text: expert | 1445 | 1489 | GPT-6.1 Sol |
| LMArena Text: hard prompts in English | 1431 | 1465 | GPT-6.1 Sol |
| LMArena Vision: diagrams | 1265 | 1306 | GPT-6.1 Sol |
| LMArena Vision: homework | 1276 | 1356 | GPT-6.1 Sol |
| Neither model has a result on the other 2 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index06
| Benchmark | GPT-5.2 | GPT-6.1 Sol | Leader |
|---|---|---|---|
| AIME | 96.9% | Not measured | Only GPT-5.2 measured |
| MGSM (Vals AI) | 94.0% | Not measured | Only GPT-5.2 measured |
| AA Math Index | 99.0 | Not measured | Only GPT-5.2 measured |
| AIME 2025 (Artificial Analysis run) | 99.0% | Not measured | Only GPT-5.2 measured |
| OTIS Mock AIME | 96.1% | 100.0% | GPT-6.1 Sol |
| FrontierMath, tiers 1 to 3 | 67.4% | 93.7% | GPT-6.1 Sol |
| FrontierMath, tier 4 | 31.7% | 100.0% | GPT-6.1 Sol |
| ProofBench | 15.0% | 99.0% | GPT-6.1 Sol |
| LMArena Text: math | 1440 | 1489 | GPT-6.1 Sol |
| LMArena Text: mathematical work | 1442 | 1513 | GPT-6.1 Sol |
| Neither model has a result on the other 3 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index15
| Benchmark | GPT-5.2 | GPT-6.1 Sol | Leader |
|---|---|---|---|
| MMLU Pro | 86.2% | Not measured | Only GPT-5.2 measured |
| LegalBench (Vals AI) | 82.8% | Not measured | Only GPT-5.2 measured |
| CaseLaw (Vals AI) | 66.0% | Not measured | Only GPT-5.2 measured |
| CorpFin (Vals AI) | 65.9% | Not measured | Only GPT-5.2 measured |
| TaxEval (Vals AI) | 75.8% | Not measured | Only GPT-5.2 measured |
| MortgageTax (Vals AI) | 67.1% | Not measured | Only GPT-5.2 measured |
| MedQA (Vals AI) | 94.1% | Not measured | Only GPT-5.2 measured |
| MedCode (Vals AI) | 49.8% | 48.8% | GPT-5.2 |
| MMLU Pro (Artificial Analysis run) | 87.4% | Not measured | Only GPT-5.2 measured |
| SimpleQA Verified | 37.1% | 73.9% | GPT-6.1 Sol |
| LMArena Text: science | 1425 | 1444 | GPT-6.1 Sol |
| LMArena Text: medicine and healthcare | 1425 | 1433 | GPT-6.1 Sol |
| LMArena Text: legal and government | 1427 | 1471 | GPT-6.1 Sol |
| LMArena Text: business and finance | 1414 | 1445 | GPT-6.1 Sol |
| Neither model has a result on the other 3 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index06
| Benchmark | GPT-5.2 | GPT-6.1 Sol | Leader |
|---|---|---|---|
| Design Arena: websites | 1196 | 1321 | GPT-6.1 Sol |
| Design Arena: UI components | 1190 | 1321 | GPT-6.1 Sol |
| Design Arena: game dev | 1196 | 1389 | GPT-6.1 Sol |
| Design Arena: data viz | 1202 | 1278 | GPT-6.1 Sol |
| Design Arena: 3D | 1079 | 1461 | GPT-6.1 Sol |
| Design Arena: all code categories | 1171 | 1345 | GPT-6.1 Sol |
| Design Arena: SVG | 1137 | Not measured | Only GPT-5.2 measured |
| Design Arena: ASCII art | 1153 | Not measured | Only GPT-5.2 measured |
| LiveBench design | Not measured | 82.1% | Only GPT-6.1 Sol measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index122
| Benchmark | GPT-5.2 | GPT-6.1 Sol | Leader |
|---|---|---|---|
| Vals Index | Not measured | 63.9% | Only GPT-6.1 Sol measured |
| LiveBench data analysis | Not measured | 82.7% | Only GPT-6.1 Sol measured |
| Public Benefits Bench (Vals AI) | Not measured | 59.3% | Only GPT-6.1 Sol measured |
| SAGE (Vals AI) | 49.3% | 46.5% | GPT-5.2 |
| MedScribe (Vals AI) | 84.4% | 86.5% | GPT-6.1 Sol |
| IFBench | 75.4% | Not measured | Only GPT-5.2 measured |
| Long context reasoning (AA-LCR) | 82.7% | 84.0% | GPT-6.1 Sol |
| DTBench | 90.9% | 96.5% | GPT-6.1 Sol |
| LMCA | 43.9% | 62.8% | GPT-6.1 Sol |
| CL-bench | 18.2% | Not measured | Only GPT-5.2 measured |
| Furniture assembly | 38.3% | 80.0% | GPT-6.1 Sol |
| LMArena Text | 1416 | 1443 | GPT-6.1 Sol |
| LMArena Text: instruction following | 1409 | 1466 | GPT-6.1 Sol |
| LMArena Text: creative writing | 1376 | 1425 | GPT-6.1 Sol |
| LMArena Vision | 1246 | 1288 | GPT-6.1 Sol |
| LMArena Text: English | 1422 | 1450 | GPT-6.1 Sol |
| LMArena Text: non-English | 1405 | 1433 | GPT-6.1 Sol |
| LMArena Text: Chinese | 1447 | 1487 | GPT-6.1 Sol |
| LMArena Text: German | 1444 | Not measured | Only GPT-5.2 measured |
| LMArena Text: Japanese | 1420 | Not measured | Only GPT-5.2 measured |
| LMArena Text: Korean | 1374 | Not measured | Only GPT-5.2 measured |
| LMArena Text: French | 1454 | 1460 | GPT-6.1 Sol |
| LMArena Text: Spanish | 1413 | 1439 | GPT-6.1 Sol |
| LMArena Text: Russian | 1415 | 1457 | GPT-6.1 Sol |
| LMArena Text: Polish | 1426 | Not measured | Only GPT-5.2 measured |
| LMArena Text: multi-turn | 1419 | 1442 | GPT-6.1 Sol |
| LMArena Text: longer queries | 1413 | 1458 | GPT-6.1 Sol |
| LMArena Text: writing and language | 1392 | 1444 | GPT-6.1 Sol |
| LMArena Text: entertainment and media | 1378 | 1426 | GPT-6.1 Sol |
| LMArena Vision: English | 1243 | 1274 | GPT-6.1 Sol |
| LMArena Vision: Chinese | 1305 | Not measured | Only GPT-5.2 measured |
| LMArena Vision: reading text | 1257 | 1289 | GPT-6.1 Sol |
| LMArena Vision: humor | 1227 | Not measured | Only GPT-5.2 measured |
| LMArena Vision: captioning | 1267 | Not measured | Only GPT-5.2 measured |
| LMArena Vision: recognizing things | 1232 | Not measured | Only GPT-5.2 measured |
| LMArena Vision: creative writing | 1225 | 1301 | GPT-6.1 Sol |
| LMArena Document | 1405 | Not measured | Only GPT-5.2 measured |
| Neither model has a result on the other 9 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | GPT-5.2 | GPT-6.1 Sol | 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
GPT-5.2$0.0088Fits in one request
GPT-6.1 Sol$0.0070Fits in one request
GPT-6.1 Sol costs less
Repository review
50K input + 3K output tokens
GPT-5.2$0.130Fits in one request
GPT-6.1 Sol$0.130Fits in one request
GPT-5.2 costs less
Agent loop
200K input + 10K output tokens
GPT-5.2$0.490Fits in one request
GPT-6.1 Sol$0.500Fits in one request
GPT-5.2 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 | OpenAI GPT-6.1 Sol |
|---|---|---|
| Maker | OpenAI | OpenAI |
| Weights | No public weights listed | No public weights listed |
| Released | December 10, 2025 | September 29, 2026 |
| Context window | 400K | 1.05M |
| Input $/Mtok | $1.75 | $2.00 |
| Output $/Mtok | $14.00 | $10.00 |
| Coding index | 58.9 | 88.7 |
| Independent measurements behind it | 2 | 2 |
| Cost per solved task | $1.91 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is GPT-5.2 better than GPT-6.1 Sol for coding?
Pick GPT-6.1 Sol for coding: it scores 88.7 on the Vibe Coding Index against 58.9 for GPT-5.2. Pick GPT-5.2 when what matters is MedCode (Vals AI) or SAGE (Vals AI), where it leads.
Which is cheaper, GPT-5.2 or GPT-6.1 Sol?
GPT-5.2 costs $1.75 per million input tokens and $14.00 per million output tokens; GPT-6.1 Sol costs $2.00 and $10.00. Cost per solved task is known for GPT-5.2 only.
Which has the bigger context window, GPT-5.2 or GPT-6.1 Sol?
GPT-6.1 Sol, with 1.05M tokens against 400K.
How many benchmarks is the GPT-5.2 vs GPT-6.1 Sol comparison based on?
62 benchmarks have a result for both models as of October 21, 2026. GPT-5.2's coding index rests on 2 independent benchmarks and GPT-6.1 Sol's on 2. Where only one model has a result, the page says so instead of naming a winner.