GPT-5.6 Sol vs Qwen3.5 397B A17B
GPT-5.6 Sol leads Qwen3.5 397B A17B on the Artificial Analysis Intelligence Index, 47.0 to 21.4, and Qwen3.5 397B A17B costs 8.5x less per output token ($2.34 vs $20.00 per million). Updated October 22, 2026; 62 benchmarks have a result for both.
GPT-5.6 Sol or Qwen3.5 397B A17B: which to pick
Only GPT-5.6 Sol has a coding score, 73.5 on the Vibe Coding Index; Qwen3.5 397B A17B has none, so the two cannot be ranked on coding.
Pick Qwen3.5 397B A17B for output tokens 8.5x cheaper ($2.34 vs $20.00 per million) and work measured by τ²-Bench Telecom or LMArena Text: science, where it leads.
- Shared results
- 62
- GPT-5.6 Sol only
- 77
- Qwen3.5 397B A17B only
- 3
- Like-for-like categories
- 8 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-5.6 Sol has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
GPT-5.6 Sol
GPT-5.6 Sol leads, 47.0 against 21.4.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only GPT-5.6 Sol has a AA Agentic Index result, 50.2.
Directional only
Reasoning
GPQA Diamond
No call
Only GPT-5.6 Sol has a GPQA Diamond result, 95.2%.
Directional only
Knowledge
MMLU Pro
No call
Only GPT-5.6 Sol has a MMLU Pro result, 89.1%.
Directional only
Web design
Design Arena: websites
GPT-5.6 Sol
GPT-5.6 Sol leads, 1321 against 1190.
Same board, both measured
Long documents
Listed context window
GPT-5.6 Sol
GPT-5.6 Sol takes 1.05M tokens against 262K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Qwen3.5 397B A17B
GPT-5.6 Sol $0.014, Qwen3.5 397B A17B $0.0016 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only GPT-5.6 Sol has the number, $1.20.
Directional only
At a glance
- Coding index
- GPT-5.6 Sol 73.5 from 6 benchmarks, Qwen3.5 397B A17B not benchmarked.
- Widest category
- Coding, 24.7 points apart on average over 8 shared benchmarks.
- Widest single gap
- FrontierMath, tiers 1 to 3: GPT-5.6 Sol 89.1% against Qwen3.5 397B A17B 31.2%, 57.9 points apart on a 0 to 100 scale.
- Output price
- GPT-5.6 Sol $20.00 against Qwen3.5 397B A17B $2.34 per million output tokens, 8.5x apart.
- Context window
- GPT-5.6 Sol takes 1.05M tokens, Qwen3.5 397B A17B 262K.
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-5.6 Sol | Bars | Qwen3.5 397B A17B | Ties |
|---|---|---|---|---|
| Coding | 8 | 0 | ||
| Intelligence | 2 | 0 | ||
| Agentic work | 0 | 1 | ||
| Reasoning | 9 | 0 | ||
| Math | 4 | 0 | ||
| Knowledge | 2 | 2 | ||
| Web design | 7 | 0 | ||
| General | 24 | 3 | ||
| All shared boards | 56 | 6 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index80
| Benchmark | GPT-5.6 Sol | Qwen3.5 397B A17B | Leader |
|---|---|---|---|
| LiveBench coding | 89.7% | Not measured | Only GPT-5.6 Sol measured |
| LMArena WebDev | 1616 | 1400 | GPT-5.6 Sol |
| AA Coding Index | 77.4 | Not measured | Only GPT-5.6 Sol measured |
| LiveCodeBench (Vals AI) | 82.6% | Not measured | Only GPT-5.6 Sol measured |
| SWE-bench Verified (Vals AI) | 96.2% | Not measured | Only GPT-5.6 Sol measured |
| Terminal-Bench 4 (Vals AI) | 37.9% | Not measured | Only GPT-5.6 Sol measured |
| Vibe Code Bench (Vals AI) | 80.5% | Not measured | Only GPT-5.6 Sol measured |
| Code migration (Vals AI) | 52.9% | Not measured | Only GPT-5.6 Sol measured |
| ProgramBench (Vals AI) | 1.5% | Not measured | Only GPT-5.6 Sol measured |
| IOI (Vals AI) | 91.2% | Not measured | Only GPT-5.6 Sol measured |
| Terminal-Bench 2.1 (Vals AI) | 85.8% | Not measured | Only GPT-5.6 Sol measured |
| Vibe Code Bench 1-100 (Vals AI) | 20.0% | Not measured | Only GPT-5.6 Sol measured |
| SkillsBench (Vals AI) | 54.1% | Not measured | Only GPT-5.6 Sol measured |
| CyberBench (Vals AI) | 76.3% | Not measured | Only GPT-5.6 Sol measured |
| SciCode | 57.8% | 44.8% | GPT-5.6 Sol |
| Terminal-Bench Hard (Artificial Analysis run) | 65.9% | 40.9% | GPT-5.6 Sol |
| FrontierCode | 47.5% | Not measured | Only GPT-5.6 Sol measured |
| FrontierSWE | 32.2% | Not measured | Only GPT-5.6 Sol measured |
| GSO-Bench | 76.5% | Not measured | Only GPT-5.6 Sol measured |
| WeirdML | 88.8% | Not measured | Only GPT-5.6 Sol measured |
| MirrorCode | 20.0% | Not measured | Only GPT-5.6 Sol measured |
| ALE-Bench | 2176.9 | 621.9 | GPT-5.6 Sol |
| LMArena Text: coding | 1496 | 1464 | GPT-5.6 Sol |
| LMArena Text: software and IT | 1489 | 1463 | GPT-5.6 Sol |
| LMArena WebDev: React | 1620 | 1388 | GPT-5.6 Sol |
| LMArena WebDev: HTML | 1618 | 1410 | GPT-5.6 Sol |
| LMArena WebDev: from an image | 1609 | Not measured | Only GPT-5.6 Sol measured |
| SciCode | 10.8% | Not measured | Only GPT-5.6 Sol measured |
| Neither model has a result on the other 11 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | GPT-5.6 Sol | Qwen3.5 397B A17B | Leader |
|---|---|---|---|
| AA Intelligence Index | 47.0 | 21.4 | GPT-5.6 Sol |
| Epoch Capabilities Index | 162.8 | 146.8 | GPT-5.6 Sol |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | GPT-5.6 Sol | Qwen3.5 397B A17B | Leader |
|---|---|---|---|
| AA Agentic Index | 50.2 | Not measured | Only GPT-5.6 Sol measured |
| Design Arena: agents building web apps | 1245 | Not measured | Only GPT-5.6 Sol measured |
| Design Arena: agents building full-stack apps | 1261 | Not measured | Only GPT-5.6 Sol measured |
| Design Arena: agents building mobile apps | 1248 | Not measured | Only GPT-5.6 Sol measured |
| Design Arena: agents building Android apps | 1240 | Not measured | Only GPT-5.6 Sol measured |
| Design Arena: agents building Godot games | 1260 | Not measured | Only GPT-5.6 Sol measured |
| Design Arena: agents building games | 1266 | Not measured | Only GPT-5.6 Sol measured |
| Design Arena: HTML slides | 1236 | Not measured | Only GPT-5.6 Sol measured |
| Design Arena: PowerPoint slides | 1217 | Not measured | Only GPT-5.6 Sol measured |
| LiveBench research | 54.5% | Not measured | Only GPT-5.6 Sol measured |
| LiveBench browser | 88.0% | Not measured | Only GPT-5.6 Sol measured |
| LiveBench long agentic | 48.5% | Not measured | Only GPT-5.6 Sol measured |
| SREBench (Vals AI) | 30.5% | Not measured | Only GPT-5.6 Sol measured |
| Finance Agent (Vals AI) | 53.8% | Not measured | Only GPT-5.6 Sol measured |
| Tax Agent Bench (Vals AI) | 31.1% | Not measured | Only GPT-5.6 Sol measured |
| Legal Research Bench (Vals AI) | 48.1% | Not measured | Only GPT-5.6 Sol measured |
| Excel Modeling Benchmark (Vals AI) | 72.3% | Not measured | Only GPT-5.6 Sol measured |
| BioMysteryBench (Vals AI) | 71.1% | Not measured | Only GPT-5.6 Sol measured |
| Terminal-Bench-Science (Vals AI) | 20.0% | Not measured | Only GPT-5.6 Sol measured |
| Vals Science Index | 41.5% | Not measured | Only GPT-5.6 Sol measured |
| Time Horizon Index (Vals AI) | 23.8% | Not measured | Only GPT-5.6 Sol measured |
| CUA-bench (Vals AI) | 8.3% | Not measured | Only GPT-5.6 Sol measured |
| τ²-Bench Telecom | 85.1% | 95.6% | Qwen3.5 397B A17B |
| APEX-Agents | Not measured | 24.9% | Only Qwen3.5 397B A17B measured |
| OSWorld 2.0 | 27.3% | Not measured | Only GPT-5.6 Sol measured |
| BALROG | 60.0% | Not measured | Only GPT-5.6 Sol measured |
| PostTrainBench | 36.2% | Not measured | Only GPT-5.6 Sol measured |
| Vending-Bench 2 | $9,619 | Not measured | Only GPT-5.6 Sol measured |
| LMArena Search | 1261 | Not measured | Only GPT-5.6 Sol measured |
| Neither model has a result on the other 12 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index90
| Benchmark | GPT-5.6 Sol | Qwen3.5 397B A17B | Leader |
|---|---|---|---|
| GPQA Diamond | 95.2% | Not measured | Only GPT-5.6 Sol measured |
| LiveBench reasoning | 91.7% | Not measured | Only GPT-5.6 Sol measured |
| MysteryMechanism (Vals AI) | 33.3% | Not measured | Only GPT-5.6 Sol measured |
| MMMU Pro (Vals AI) | 88.8% | Not measured | Only GPT-5.6 Sol measured |
| GPQA Diamond (Artificial Analysis run) | 94.1% | 89.3% | GPT-5.6 Sol |
| Humanity's Last Exam (Artificial Analysis run) | 49.5% | 29.0% | GPT-5.6 Sol |
| GPQA Diamond (Epoch AI run) | 93.5% | 86.4% | GPT-5.6 Sol |
| ARC-AGI | 97.5% | Not measured | Only GPT-5.6 Sol measured |
| ARC-AGI-2 | 92.5% | Not measured | Only GPT-5.6 Sol measured |
| SimpleBench | 64.8% | Not measured | Only GPT-5.6 Sol measured |
| Chess puzzles | 59.0% | Not measured | Only GPT-5.6 Sol measured |
| Mystery game puzzles | 58.0% | 18.0% | GPT-5.6 Sol |
| EBR-bench | 44.8% | Not measured | Only GPT-5.6 Sol measured |
| LMArena Text: hard prompts | 1483 | 1448 | GPT-5.6 Sol |
| LMArena Text: expert | 1517 | 1462 | GPT-5.6 Sol |
| LMArena Text: hard prompts in English | 1477 | 1450 | GPT-5.6 Sol |
| LMArena Vision: diagrams | 1305 | 1277 | GPT-5.6 Sol |
| LMArena Vision: homework | 1305 | 1292 | GPT-5.6 Sol |
| Kagi LLM Benchmark | 67.0% | Not measured | Only GPT-5.6 Sol measured |
| Neither model has a result on the other 3 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index40
| Benchmark | GPT-5.6 Sol | Qwen3.5 397B A17B | Leader |
|---|---|---|---|
| OTIS Mock AIME | 100.0% | 88.9% | GPT-5.6 Sol |
| FrontierMath, tiers 1 to 3 | 89.1% | 31.2% | GPT-5.6 Sol |
| FrontierMath, tier 4 | 82.9% | Not measured | Only GPT-5.6 Sol measured |
| ProofBench | 83.0% | Not measured | Only GPT-5.6 Sol measured |
| LMArena Text: math | 1473 | 1453 | GPT-5.6 Sol |
| LMArena Text: mathematical work | 1496 | 1454 | GPT-5.6 Sol |
| Neither model has a result on the other 7 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index22
| Benchmark | GPT-5.6 Sol | Qwen3.5 397B A17B | Leader |
|---|---|---|---|
| MMLU Pro | 89.1% | Not measured | Only GPT-5.6 Sol measured |
| LegalBench (Vals AI) | 87.0% | Not measured | Only GPT-5.6 Sol measured |
| CorpFin (Vals AI) | 64.4% | Not measured | Only GPT-5.6 Sol measured |
| TaxEval (Vals AI) | 74.8% | Not measured | Only GPT-5.6 Sol measured |
| MortgageTax (Vals AI) | 67.3% | Not measured | Only GPT-5.6 Sol measured |
| MedCode (Vals AI) | 44.0% | Not measured | Only GPT-5.6 Sol measured |
| SimpleQA Verified | 69.7% | Not measured | Only GPT-5.6 Sol measured |
| LMArena Text: science | 1455 | 1457 | Qwen3.5 397B A17B |
| LMArena Text: medicine and healthcare | 1431 | 1455 | Qwen3.5 397B A17B |
| LMArena Text: legal and government | 1476 | 1441 | GPT-5.6 Sol |
| LMArena Text: business and finance | 1457 | 1435 | GPT-5.6 Sol |
| FACTS Benchmark Suite | 52.4% | Not measured | Only GPT-5.6 Sol measured |
| Neither model has a result on the other 6 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index70
| Benchmark | GPT-5.6 Sol | Qwen3.5 397B A17B | Leader |
|---|---|---|---|
| Design Arena: websites | 1321 | 1190 | GPT-5.6 Sol |
| Design Arena: UI components | 1335 | 1161 | GPT-5.6 Sol |
| Design Arena: game dev | 1367 | 1140 | GPT-5.6 Sol |
| Design Arena: data viz | 1323 | 1175 | GPT-5.6 Sol |
| Design Arena: 3D | 1384 | 1160 | GPT-5.6 Sol |
| Design Arena: all code categories | 1341 | 1179 | GPT-5.6 Sol |
| Design Arena: SVG | 1315 | 1139 | GPT-5.6 Sol |
| Design Arena: ASCII art | 1304 | Not measured | Only GPT-5.6 Sol measured |
| LiveBench design | 81.6% | Not measured | Only GPT-5.6 Sol measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index243
| Benchmark | GPT-5.6 Sol | Qwen3.5 397B A17B | Leader |
|---|---|---|---|
| LiveBench data analysis | 79.8% | Not measured | Only GPT-5.6 Sol measured |
| Public Benefits Bench (Vals AI) | 66.5% | Not measured | Only GPT-5.6 Sol measured |
| SAGE (Vals AI) | 52.6% | Not measured | Only GPT-5.6 Sol measured |
| MedScribe (Vals AI) | 85.2% | Not measured | Only GPT-5.6 Sol measured |
| Vals Multimodal Index | 72.6% | Not measured | Only GPT-5.6 Sol measured |
| IFBench | 72.7% | 78.8% | Qwen3.5 397B A17B |
| Long context reasoning (AA-LCR) | 84.0% | 77.3% | GPT-5.6 Sol |
| DTBench | 95.7% | 87.5% | GPT-5.6 Sol |
| LMCA | 58.5% | 37.9% | GPT-5.6 Sol |
| Surface Evolver Bench | 93.1% | Not measured | Only GPT-5.6 Sol measured |
| Furniture assembly | 56.7% | Not measured | Only GPT-5.6 Sol measured |
| LMArena Text | 1457 | 1437 | GPT-5.6 Sol |
| LMArena Text: instruction following | 1483 | 1424 | GPT-5.6 Sol |
| LMArena Text: creative writing | 1449 | 1402 | GPT-5.6 Sol |
| LMArena Vision | 1281 | 1262 | GPT-5.6 Sol |
| LMArena Text: English | 1457 | 1443 | GPT-5.6 Sol |
| LMArena Text: non-English | 1451 | 1430 | GPT-5.6 Sol |
| LMArena Text: Chinese | 1525 | 1498 | GPT-5.6 Sol |
| LMArena Text: German | 1476 | 1445 | GPT-5.6 Sol |
| LMArena Text: Japanese | 1472 | 1429 | GPT-5.6 Sol |
| LMArena Text: Korean | 1442 | 1384 | GPT-5.6 Sol |
| LMArena Text: French | 1472 | 1462 | GPT-5.6 Sol |
| LMArena Text: Spanish | 1443 | 1441 | GPT-5.6 Sol |
| LMArena Text: Russian | 1471 | 1430 | GPT-5.6 Sol |
| LMArena Text: Polish | 1459 | 1447 | GPT-5.6 Sol |
| LMArena Text: multi-turn | 1461 | 1446 | GPT-5.6 Sol |
| LMArena Text: longer queries | 1481 | 1441 | GPT-5.6 Sol |
| LMArena Text: writing and language | 1462 | 1415 | GPT-5.6 Sol |
| LMArena Text: entertainment and media | 1450 | 1399 | GPT-5.6 Sol |
| LMArena Vision: English | 1280 | 1268 | GPT-5.6 Sol |
| LMArena Vision: Chinese | 1313 | 1314 | Qwen3.5 397B A17B |
| LMArena Vision: reading text | 1290 | 1271 | GPT-5.6 Sol |
| LMArena Vision: humor | 1246 | 1262 | Qwen3.5 397B A17B |
| LMArena Vision: captioning | Not measured | 1230 | Only Qwen3.5 397B A17B measured |
| LMArena Vision: recognizing things | Not measured | 1265 | Only Qwen3.5 397B A17B measured |
| LMArena Vision: creative writing | 1283 | 1260 | GPT-5.6 Sol |
| LMArena Document | 1483 | Not measured | Only GPT-5.6 Sol 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-5.6 Sol | Qwen3.5 397B A17B | 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
Qwen3.5 397B A17B costs less
Repository review
50K input + 3K output tokens
Qwen3.5 397B A17B costs less
Agent loop
200K input + 10K output tokens
Qwen3.5 397B A17B 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-5.6 Sol | Qwen3.5 397B A17B |
|---|---|---|
| Maker | OpenAI | Qwen |
| Weights | No public weights listed | Open weights |
| Released | July 9, 2026 | February 16, 2026 |
| Context window | 1.05M | 262K |
| Input $/Mtok | $4.00 | $0.39 |
| Output $/Mtok | $20.00 | $2.34 |
| Coding index | 73.5 | Not benchmarked |
| Independent measurements behind it | 6 | 1 |
| Cost per solved task | $1.20 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is GPT-5.6 Sol better than Qwen3.5 397B A17B for coding?
Only GPT-5.6 Sol has a coding score, 73.5 on the Vibe Coding Index; Qwen3.5 397B A17B has none, so the two cannot be ranked on coding. Pick Qwen3.5 397B A17B for output tokens 8.5x cheaper ($2.34 vs $20.00 per million) and work measured by τ²-Bench Telecom or LMArena Text: science, where it leads.
Which is cheaper, GPT-5.6 Sol or Qwen3.5 397B A17B?
GPT-5.6 Sol costs $4.00 per million input tokens and $20.00 per million output tokens; Qwen3.5 397B A17B costs $0.39 and $2.34. Cost per solved task is known for GPT-5.6 Sol only.
Which has the bigger context window, GPT-5.6 Sol or Qwen3.5 397B A17B?
GPT-5.6 Sol, with 1.05M tokens against 262K.
How many benchmarks is the GPT-5.6 Sol vs Qwen3.5 397B A17B comparison based on?
62 benchmarks have a result for both models as of October 22, 2026. GPT-5.6 Sol's coding index rests on 6 independent benchmarks and Qwen3.5 397B A17B's on 1. Where only one model has a result, the page says so instead of naming a winner.