GPT-5.1 vs Qwen2.5 VL 72B Instruct
Qwen2.5 VL 72B Instruct costs 10.0x less per output token than GPT-5.1 ($1.00 vs $10.00 per million). Updated October 22, 2026; 3 benchmarks have a result for both.
GPT-5.1 or Qwen2.5 VL 72B Instruct: which to pick
Only GPT-5.1 has a coding score, 40.4 on the Vibe Coding Index; Qwen2.5 VL 72B Instruct has none, so the two cannot be ranked on coding.
Pick Qwen2.5 VL 72B Instruct for output tokens 10.0x cheaper ($1.00 vs $10.00 per million).
- Shared results
- 3
- GPT-5.1 only
- 98
- Qwen2.5 VL 72B Instruct only
- 3
- Like-for-like categories
- 1 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
No call
Only GPT-5.1 has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
No call
Only GPT-5.1 has a AA Intelligence Index result, 24.7.
Directional only
Reasoning
GPQA Diamond
No call
Only GPT-5.1 has a GPQA Diamond result, 86.6%.
Directional only
Math
AIME
No call
Only GPT-5.1 has a AIME result, 93.3%.
Directional only
Knowledge
MMLU Pro
No call
Only GPT-5.1 has a MMLU Pro result, 86.4%.
Directional only
Web design
Design Arena: websites
No call
Only GPT-5.1 has a Design Arena: websites result, 1187.
Directional only
Long documents
Listed context window
GPT-5.1
GPT-5.1 takes 400K tokens against 128K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Qwen2.5 VL 72B Instruct
GPT-5.1 $0.0063, Qwen2.5 VL 72B Instruct $0.0013 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only GPT-5.1 has the number, $1.02.
Directional only
At a glance
- Coding index
- GPT-5.1 40.4 from 3 benchmarks, Qwen2.5 VL 72B Instruct not benchmarked.
- Widest category
- General, 24.6 points apart on average over 3 shared benchmarks.
- Widest single gap
- LMArena Vision: Chinese: GPT-5.1 1270 against Qwen2.5 VL 72B Instruct 1107, 27.2 points apart on a 0 to 100 scale.
- Output price
- GPT-5.1 $10.00 against Qwen2.5 VL 72B Instruct $1.00 per million output tokens, 10.0x apart.
- Context window
- GPT-5.1 takes 400K tokens, Qwen2.5 VL 72B Instruct 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.
| Category | GPT-5.1 | Bars | Qwen2.5 VL 72B Instruct | Ties |
|---|---|---|---|---|
| General | 3 | 0 | ||
| All shared boards | 3 | 0 |
CodingDirectional onlySix of these boards feed the Vibe Coding Index16 boards, none shared
| Benchmark | GPT-5.1 | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| SWE-bench Verified | 66.0% | Not measured | Only GPT-5.1 measured |
| LMArena WebDev | 1396 | Not measured | Only GPT-5.1 measured |
| AA Coding Index | 49.4 | Not measured | Only GPT-5.1 measured |
| LiveCodeBench (Vals AI) | 86.5% | Not measured | Only GPT-5.1 measured |
| SWE-bench Verified (Vals AI) | 69.8% | Not measured | Only GPT-5.1 measured |
| Vibe Code Bench (Vals AI) | 24.6% | Not measured | Only GPT-5.1 measured |
| LiveCodeBench (Artificial Analysis run) | 86.8% | Not measured | Only GPT-5.1 measured |
| Terminal-Bench Hard (Artificial Analysis run) | 45.5% | Not measured | Only GPT-5.1 measured |
| Terminal-Bench | 47.6% | Not measured | Only GPT-5.1 measured |
| GSO-Bench | 13.7% | Not measured | Only GPT-5.1 measured |
| WeirdML | 60.8% | Not measured | Only GPT-5.1 measured |
| LMArena Text: coding | 1453 | Not measured | Only GPT-5.1 measured |
| LMArena Text: software and IT | 1454 | Not measured | Only GPT-5.1 measured |
| LMArena WebDev: React | 1307 | Not measured | Only GPT-5.1 measured |
| LMArena WebDev: HTML | 1405 | Not measured | Only GPT-5.1 measured |
| LMArena WebDev: from an image | 1421 | Not measured | Only GPT-5.1 measured |
| Neither model has a result on the other 23 boards here | |||
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
| Benchmark | GPT-5.1 | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| AA Intelligence Index | 24.7 | Not measured | Only GPT-5.1 measured |
| Epoch Capabilities Index | 149.8 | Not measured | Only GPT-5.1 measured |
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index4 boards, none shared
| Benchmark | GPT-5.1 | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| τ²-Bench Telecom | 81.9% | Not measured | Only GPT-5.1 measured |
| DeepResearch Bench | 42.8% | Not measured | Only GPT-5.1 measured |
| OSWorld | Not measured | 5.0% | Only Qwen2.5 VL 72B Instruct measured |
| Vending-Bench 2 | $1,473 | Not measured | Only GPT-5.1 measured |
| Neither model has a result on the other 37 boards here | |||
ReasoningDirectional onlyPublic boards, shown as published, kept out of the index17 boards, none shared
| Benchmark | GPT-5.1 | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| GPQA Diamond | 86.6% | Not measured | Only GPT-5.1 measured |
| MMMU Pro (Vals AI) | 83.2% | Not measured | Only GPT-5.1 measured |
| GPQA Diamond (Artificial Analysis run) | 87.3% | Not measured | Only GPT-5.1 measured |
| Humanity's Last Exam (Artificial Analysis run) | 28.5% | Not measured | Only GPT-5.1 measured |
| GPQA Diamond (Epoch AI run) | 87.6% | Not measured | Only GPT-5.1 measured |
| ARC-AGI | 72.8% | Not measured | Only GPT-5.1 measured |
| ARC-AGI-2 | 17.6% | Not measured | Only GPT-5.1 measured |
| SimpleBench | 53.2% | Not measured | Only GPT-5.1 measured |
| Chess puzzles | 42.0% | Not measured | Only GPT-5.1 measured |
| Mystery game puzzles | 19.0% | Not measured | Only GPT-5.1 measured |
| VPCT | 58.7% | Not measured | Only GPT-5.1 measured |
| LMArena Text: hard prompts | 1457 | Not measured | Only GPT-5.1 measured |
| LMArena Text: expert | 1470 | Not measured | Only GPT-5.1 measured |
| LMArena Text: hard prompts in English | 1462 | Not measured | Only GPT-5.1 measured |
| LMArena Vision: diagrams | 1264 | Not measured | Only GPT-5.1 measured |
| LMArena Vision: homework | 1272 | Not measured | Only GPT-5.1 measured |
| Kagi LLM Benchmark | Not measured | 36.0% | Only Qwen2.5 VL 72B Instruct measured |
| Neither model has a result on the other 5 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index7 boards, none shared
| Benchmark | GPT-5.1 | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| AIME | 93.3% | Not measured | Only GPT-5.1 measured |
| MGSM (Vals AI) | 93.0% | Not measured | Only GPT-5.1 measured |
| AA Math Index | 94.0 | Not measured | Only GPT-5.1 measured |
| AIME 2025 (Artificial Analysis run) | 94.0% | Not measured | Only GPT-5.1 measured |
| OTIS Mock AIME | 88.6% | Not measured | Only GPT-5.1 measured |
| LMArena Text: math | 1446 | Not measured | Only GPT-5.1 measured |
| LMArena Text: mathematical work | 1449 | Not measured | Only GPT-5.1 measured |
| Neither model has a result on the other 6 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index16 boards, none shared
| Benchmark | GPT-5.1 | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| MMLU Pro | 86.4% | Not measured | Only GPT-5.1 measured |
| LegalBench (Vals AI) | 85.7% | Not measured | Only GPT-5.1 measured |
| CaseLaw (Vals AI) | 73.4% | Not measured | Only GPT-5.1 measured |
| CorpFin (Vals AI) | 63.8% | Not measured | Only GPT-5.1 measured |
| TaxEval (Vals AI) | 74.9% | Not measured | Only GPT-5.1 measured |
| MortgageTax (Vals AI) | 61.4% | Not measured | Only GPT-5.1 measured |
| MedQA (Vals AI) | 96.4% | Not measured | Only GPT-5.1 measured |
| MedCode (Vals AI) | 52.7% | Not measured | Only GPT-5.1 measured |
| MMLU Pro (Artificial Analysis run) | 87.0% | Not measured | Only GPT-5.1 measured |
| SimpleQA Verified | 48.0% | Not measured | Only GPT-5.1 measured |
| GeoBench | Not measured | 62.0% | Only Qwen2.5 VL 72B Instruct measured |
| LMArena Text: science | 1459 | Not measured | Only GPT-5.1 measured |
| LMArena Text: medicine and healthcare | 1456 | Not measured | Only GPT-5.1 measured |
| LMArena Text: legal and government | 1459 | Not measured | Only GPT-5.1 measured |
| LMArena Text: business and finance | 1431 | Not measured | Only GPT-5.1 measured |
| FACTS Benchmark Suite | 45.0% | Not measured | Only GPT-5.1 measured |
| Neither model has a result on the other 2 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
| Benchmark | GPT-5.1 | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| Design Arena: websites | 1187 | Not measured | Only GPT-5.1 measured |
| Design Arena: UI components | 1163 | Not measured | Only GPT-5.1 measured |
| Design Arena: game dev | 1178 | Not measured | Only GPT-5.1 measured |
| Design Arena: data viz | 1204 | Not measured | Only GPT-5.1 measured |
| Design Arena: 3D | 1060 | Not measured | Only GPT-5.1 measured |
| Design Arena: all code categories | 1170 | Not measured | Only GPT-5.1 measured |
| Design Arena: SVG | 1149 | Not measured | Only GPT-5.1 measured |
| Design Arena: ASCII art | 1119 | Not measured | Only GPT-5.1 measured |
| Neither model has a result on the other 1 board here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index30
| Benchmark | GPT-5.1 | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| SAGE (Vals AI) | 43.2% | Not measured | Only GPT-5.1 measured |
| MedScribe (Vals AI) | 88.1% | Not measured | Only GPT-5.1 measured |
| IFBench | 72.9% | Not measured | Only GPT-5.1 measured |
| Long context reasoning (AA-LCR) | 80.0% | Not measured | Only GPT-5.1 measured |
| DTBench | 90.1% | Not measured | Only GPT-5.1 measured |
| LMCA | 43.9% | Not measured | Only GPT-5.1 measured |
| CL-bench | 23.7% | Not measured | Only GPT-5.1 measured |
| CL-bench Life | 17.3% | Not measured | Only GPT-5.1 measured |
| LMArena Text | 1443 | Not measured | Only GPT-5.1 measured |
| LMArena Text: instruction following | 1443 | Not measured | Only GPT-5.1 measured |
| LMArena Text: creative writing | 1427 | Not measured | Only GPT-5.1 measured |
| LMArena Vision | 1250 | 1107 | GPT-5.1 |
| LMArena Text: English | 1451 | Not measured | Only GPT-5.1 measured |
| LMArena Text: non-English | 1431 | Not measured | Only GPT-5.1 measured |
| LMArena Text: Chinese | 1495 | Not measured | Only GPT-5.1 measured |
| LMArena Text: German | 1438 | Not measured | Only GPT-5.1 measured |
| LMArena Text: Japanese | 1453 | Not measured | Only GPT-5.1 measured |
| LMArena Text: Korean | 1401 | Not measured | Only GPT-5.1 measured |
| LMArena Text: French | 1450 | Not measured | Only GPT-5.1 measured |
| LMArena Text: Spanish | 1433 | Not measured | Only GPT-5.1 measured |
| LMArena Text: Russian | 1435 | Not measured | Only GPT-5.1 measured |
| LMArena Text: Polish | 1440 | Not measured | Only GPT-5.1 measured |
| LMArena Text: multi-turn | 1450 | Not measured | Only GPT-5.1 measured |
| LMArena Text: longer queries | 1448 | Not measured | Only GPT-5.1 measured |
| LMArena Text: writing and language | 1428 | Not measured | Only GPT-5.1 measured |
| LMArena Text: entertainment and media | 1415 | Not measured | Only GPT-5.1 measured |
| LMArena Vision: English | 1252 | 1115 | GPT-5.1 |
| LMArena Vision: Chinese | 1270 | 1107 | GPT-5.1 |
| LMArena Vision: reading text | 1255 | Not measured | Only GPT-5.1 measured |
| LMArena Vision: humor | 1263 | Not measured | Only GPT-5.1 measured |
| LMArena Vision: captioning | 1252 | Not measured | Only GPT-5.1 measured |
| LMArena Vision: recognizing things | 1232 | Not measured | Only GPT-5.1 measured |
| LMArena Vision: creative writing | 1266 | Not measured | Only GPT-5.1 measured |
| LMArena Document | 1403 | Not measured | Only GPT-5.1 measured |
| Neither model has a result on the other 12 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | GPT-5.1 | Qwen2.5 VL 72B Instruct | 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
Qwen2.5 VL 72B Instruct costs less
Repository review
50K input + 3K output tokens
Qwen2.5 VL 72B Instruct costs less
Agent loop
200K input + 10K output tokens
Qwen2.5 VL 72B Instruct 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-5.1 | Qwen2.5 VL 72B Instruct |
|---|---|---|
| Maker | OpenAI | Qwen |
| Weights | No public weights listed | Open weights |
| Released | November 13, 2025 | February 1, 2025 |
| Context window | 400K | 128K |
| Input $/Mtok | $1.25 | $0.80 |
| Output $/Mtok | $10.00 | $1.00 |
| Coding index | 40.4 | Not benchmarked |
| Independent measurements behind it | 3 | 0 |
| Cost per solved task | $1.02 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is GPT-5.1 better than Qwen2.5 VL 72B Instruct for coding?
Only GPT-5.1 has a coding score, 40.4 on the Vibe Coding Index; Qwen2.5 VL 72B Instruct has none, so the two cannot be ranked on coding. Pick Qwen2.5 VL 72B Instruct for output tokens 10.0x cheaper ($1.00 vs $10.00 per million).
Which is cheaper, GPT-5.1 or Qwen2.5 VL 72B Instruct?
GPT-5.1 costs $1.25 per million input tokens and $10.00 per million output tokens; Qwen2.5 VL 72B Instruct costs $0.80 and $1.00. Cost per solved task is known for GPT-5.1 only.
Which has the bigger context window, GPT-5.1 or Qwen2.5 VL 72B Instruct?
GPT-5.1, with 400K tokens against 128K.
How many benchmarks is the GPT-5.1 vs Qwen2.5 VL 72B Instruct comparison based on?
3 benchmarks have a result for both models as of October 22, 2026. GPT-5.1's coding index rests on 3 independent benchmarks and Qwen2.5 VL 72B Instruct's on 0. Where only one model has a result, the page says so instead of naming a winner.