GPT-5.2-Codex vs Qwen3.6 Plus
GPT-5.2-Codex leads Qwen3.6 Plus on coding, 45.4 to 44.1 on the Vibe Coding Index, and Qwen3.6 Plus costs 7.2x less per output token ($1.95 vs $14.00 per million). Updated October 22, 2026; 14 benchmarks have a result for both.
GPT-5.2-Codex or Qwen3.6 Plus: which to pick
GPT-5.2-Codex and Qwen3.6 Plus are level for coding, 45.4 against 44.1 on the Vibe Coding Index, so price and context decide.
Pick Qwen3.6 Plus for output tokens 7.2x cheaper ($1.95 vs $14.00 per million) and a larger context window (1M against 400K tokens). Pick GPT-5.2-Codex for a lower cost per solved task ($0.43 against $1.19) and work measured by LiveCodeBench (Vals AI) or Vibe Code Bench (Vals AI), where it leads.
- Shared results
- 14
- GPT-5.2-Codex only
- 8
- Qwen3.6 Plus only
- 69
- 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
No call
Level: 45.4 against 44.1, inside 2 points.
3 benchmarks behind each score
Intelligence
AA Intelligence Index
GPT-5.2-Codex
GPT-5.2-Codex leads, 28.5 against 27.0.
Same board, both measured
Reasoning
GPQA Diamond
No call
Only Qwen3.6 Plus has a GPQA Diamond result, 87.4%.
Directional only
Math
AIME
No call
Only Qwen3.6 Plus has a AIME result, 94.6%.
Directional only
Knowledge
MMLU Pro
No call
Only Qwen3.6 Plus has a MMLU Pro result, 87.7%.
Directional only
Web design
Design Arena: websites
No call
Only Qwen3.6 Plus has a Design Arena: websites result, 1243.
Directional only
Long documents
Listed context window
Qwen3.6 Plus
Qwen3.6 Plus takes 1M tokens against 400K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Qwen3.6 Plus
GPT-5.2-Codex $0.0088, Qwen3.6 Plus $0.0013 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
GPT-5.2-Codex
GPT-5.2-Codex $0.43, Qwen3.6 Plus $1.19.
Published attempt cost and pass rate
At a glance
- Coding index
- GPT-5.2-Codex 45.4 from 3 benchmarks, Qwen3.6 Plus 44.1 from 3 benchmarks.
- Widest category
- Coding, 12.6 points apart on average over 8 shared benchmarks.
- Widest single gap
- ALE-Bench: GPT-5.2-Codex 1299.9 against Qwen3.6 Plus 670.1, 21.0 points apart on a 0 to 100 scale.
- Output price
- GPT-5.2-Codex $14.00 against Qwen3.6 Plus $1.95 per million output tokens, 7.2x apart.
- Context window
- GPT-5.2-Codex takes 400K tokens, Qwen3.6 Plus 1M.
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 | Qwen3.6 Plus | Ties |
|---|---|---|---|---|
| Coding | 3 | 5 | ||
| Intelligence | 1 | 0 | ||
| Agentic work | 0 | 1 | ||
| Reasoning | 2 | 0 | ||
| General | 2 | 0 | ||
| All shared boards | 8 | 6 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index35
| Benchmark | GPT-5.2-Codex | Qwen3.6 Plus | Leader |
|---|---|---|---|
| SWE-bench Verified | 72.8% | Not measured | Only GPT-5.2-Codex measured |
| SWE-bench Multilingual | 66.3% | Not measured | Only GPT-5.2-Codex measured |
| LMArena WebDev | 1339 | 1461 | Qwen3.6 Plus |
| AA Coding Index | Not measured | 54.5 | Only Qwen3.6 Plus measured |
| LiveCodeBench (Vals AI) | 88.0% | 86.0% | GPT-5.2-Codex |
| SWE-bench Verified (Vals AI) | 72.4% | 73.4% | Qwen3.6 Plus |
| Vibe Code Bench (Vals AI) | 37.9% | 25.6% | GPT-5.2-Codex |
| Code migration (Vals AI) | Not measured | 11.1% | Only Qwen3.6 Plus measured |
| ProgramBench (Vals AI) | Not measured | 0.0% | Only Qwen3.6 Plus measured |
| Terminal-Bench 2.1 (Vals AI) | Not measured | 53.2% | Only Qwen3.6 Plus measured |
| Terminal-Bench Hard (Artificial Analysis run) | 37.1% | 43.9% | Qwen3.6 Plus |
| Terminal-Bench | 66.5% | Not measured | Only GPT-5.2-Codex measured |
| ALE-Bench | 1299.9 | 670.1 | GPT-5.2-Codex |
| LMArena Text: coding | Not measured | 1467 | Only Qwen3.6 Plus measured |
| LMArena Text: software and IT | Not measured | 1462 | Only Qwen3.6 Plus measured |
| LMArena WebDev: React | 1329 | 1454 | Qwen3.6 Plus |
| LMArena WebDev: HTML | 1348 | 1463 | Qwen3.6 Plus |
| LMArena WebDev: from an image | Not measured | 1472 | Only Qwen3.6 Plus measured |
| Neither model has a result on the other 21 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | GPT-5.2-Codex | Qwen3.6 Plus | Leader |
|---|---|---|---|
| AA Intelligence Index | 28.5 | 27.0 | GPT-5.2-Codex |
| Epoch Capabilities Index | Not measured | 147.3 | Only Qwen3.6 Plus measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | GPT-5.2-Codex | Qwen3.6 Plus | 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 |
| Finance Agent (Vals AI) | Not measured | 40.9% | Only Qwen3.6 Plus measured |
| Tax Agent Bench (Vals AI) | Not measured | 7.7% | Only Qwen3.6 Plus measured |
| Legal Research Bench (Vals AI) | Not measured | 14.9% | Only Qwen3.6 Plus measured |
| Excel Modeling Benchmark (Vals AI) | Not measured | 32.9% | Only Qwen3.6 Plus measured |
| τ²-Bench Telecom | 92.1% | 97.7% | Qwen3.6 Plus |
| Vending-Bench 2 | Not measured | $5,115 | Only Qwen3.6 Plus measured |
| Neither model has a result on the other 30 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | GPT-5.2-Codex | Qwen3.6 Plus | Leader |
|---|---|---|---|
| GPQA Diamond | Not measured | 87.4% | Only Qwen3.6 Plus measured |
| MMMU Pro (Vals AI) | Not measured | 84.2% | Only Qwen3.6 Plus measured |
| GPQA Diamond (Artificial Analysis run) | 89.9% | 88.2% | GPT-5.2-Codex |
| Humanity's Last Exam (Artificial Analysis run) | 35.7% | 27.8% | GPT-5.2-Codex |
| GPQA Diamond (Epoch AI run) | Not measured | 88.4% | Only Qwen3.6 Plus measured |
| Chess puzzles | Not measured | 14.0% | Only Qwen3.6 Plus measured |
| Mystery game puzzles | Not measured | 12.0% | Only Qwen3.6 Plus measured |
| LMArena Text: hard prompts | Not measured | 1448 | Only Qwen3.6 Plus measured |
| LMArena Text: expert | Not measured | 1454 | Only Qwen3.6 Plus measured |
| LMArena Text: hard prompts in English | Not measured | 1455 | Only Qwen3.6 Plus measured |
| Neither model has a result on the other 12 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index5 boards, none shared
| Benchmark | GPT-5.2-Codex | Qwen3.6 Plus | Leader |
|---|---|---|---|
| AIME | Not measured | 94.6% | Only Qwen3.6 Plus measured |
| OTIS Mock AIME | Not measured | 93.3% | Only Qwen3.6 Plus measured |
| FrontierMath, tiers 1 to 3 | Not measured | 38.2% | Only Qwen3.6 Plus measured |
| LMArena Text: math | Not measured | 1449 | Only Qwen3.6 Plus measured |
| LMArena Text: mathematical work | Not measured | 1448 | Only Qwen3.6 Plus measured |
| Neither model has a result on the other 8 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index12 boards, none shared
| Benchmark | GPT-5.2-Codex | Qwen3.6 Plus | Leader |
|---|---|---|---|
| MMLU Pro | Not measured | 87.7% | Only Qwen3.6 Plus measured |
| LegalBench (Vals AI) | Not measured | 84.2% | Only Qwen3.6 Plus measured |
| CaseLaw (Vals AI) | Not measured | 51.5% | Only Qwen3.6 Plus measured |
| CorpFin (Vals AI) | Not measured | 61.9% | Only Qwen3.6 Plus measured |
| TaxEval (Vals AI) | Not measured | 74.7% | Only Qwen3.6 Plus measured |
| MortgageTax (Vals AI) | Not measured | 68.0% | Only Qwen3.6 Plus measured |
| MedCode (Vals AI) | Not measured | 36.9% | Only Qwen3.6 Plus measured |
| SimpleQA Verified | Not measured | 44.1% | Only Qwen3.6 Plus measured |
| LMArena Text: science | Not measured | 1439 | Only Qwen3.6 Plus measured |
| LMArena Text: medicine and healthcare | Not measured | 1442 | Only Qwen3.6 Plus measured |
| LMArena Text: legal and government | Not measured | 1437 | Only Qwen3.6 Plus measured |
| LMArena Text: business and finance | Not measured | 1430 | Only Qwen3.6 Plus measured |
| Neither model has a result on the other 6 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
| Benchmark | GPT-5.2-Codex | Qwen3.6 Plus | Leader |
|---|---|---|---|
| Design Arena: websites | Not measured | 1243 | Only Qwen3.6 Plus measured |
| Design Arena: UI components | Not measured | 1230 | Only Qwen3.6 Plus measured |
| Design Arena: game dev | Not measured | 1211 | Only Qwen3.6 Plus measured |
| Design Arena: data viz | Not measured | 1228 | Only Qwen3.6 Plus measured |
| Design Arena: 3D | Not measured | 1202 | Only Qwen3.6 Plus measured |
| Design Arena: all code categories | Not measured | 1236 | Only Qwen3.6 Plus measured |
| Design Arena: SVG | Not measured | 1161 | Only Qwen3.6 Plus measured |
| Design Arena: ASCII art | Not measured | 1113 | Only Qwen3.6 Plus measured |
| Neither model has a result on the other 1 board here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | GPT-5.2-Codex | Qwen3.6 Plus | Leader |
|---|---|---|---|
| SAGE (Vals AI) | Not measured | 44.9% | Only Qwen3.6 Plus measured |
| MedScribe (Vals AI) | Not measured | 77.0% | Only Qwen3.6 Plus measured |
| Vals Multimodal Index | Not measured | 51.5% | Only Qwen3.6 Plus measured |
| IFBench | 77.6% | 75.2% | GPT-5.2-Codex |
| Long context reasoning (AA-LCR) | 82.3% | 78.3% | GPT-5.2-Codex |
| DTBench | Not measured | 81.9% | Only Qwen3.6 Plus measured |
| LMCA | Not measured | 33.1% | Only Qwen3.6 Plus measured |
| CL-bench | Not measured | 20.3% | Only Qwen3.6 Plus measured |
| LMArena Text | Not measured | 1436 | Only Qwen3.6 Plus measured |
| LMArena Text: instruction following | Not measured | 1424 | Only Qwen3.6 Plus measured |
| LMArena Text: creative writing | Not measured | 1404 | Only Qwen3.6 Plus measured |
| LMArena Text: English | Not measured | 1445 | Only Qwen3.6 Plus measured |
| LMArena Text: non-English | Not measured | 1424 | Only Qwen3.6 Plus measured |
| LMArena Text: Chinese | Not measured | 1477 | Only Qwen3.6 Plus measured |
| LMArena Text: German | Not measured | 1451 | Only Qwen3.6 Plus measured |
| LMArena Text: Japanese | Not measured | 1390 | Only Qwen3.6 Plus measured |
| LMArena Text: Korean | Not measured | 1379 | Only Qwen3.6 Plus measured |
| LMArena Text: French | Not measured | 1455 | Only Qwen3.6 Plus measured |
| LMArena Text: Spanish | Not measured | 1432 | Only Qwen3.6 Plus measured |
| LMArena Text: Russian | Not measured | 1434 | Only Qwen3.6 Plus measured |
| LMArena Text: Polish | Not measured | 1433 | Only Qwen3.6 Plus measured |
| LMArena Text: multi-turn | Not measured | 1438 | Only Qwen3.6 Plus measured |
| LMArena Text: longer queries | Not measured | 1439 | Only Qwen3.6 Plus measured |
| LMArena Text: writing and language | Not measured | 1418 | Only Qwen3.6 Plus measured |
| LMArena Text: entertainment and media | Not measured | 1402 | Only Qwen3.6 Plus measured |
| Neither model has a result on the other 21 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | GPT-5.2-Codex | Qwen3.6 Plus | 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
Qwen3.6 Plus costs less
Repository review
50K input + 3K output tokens
Qwen3.6 Plus costs less
Agent loop
200K input + 10K output tokens
Qwen3.6 Plus 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 | Qwen3.6 Plus |
|---|---|---|
| Maker | OpenAI | Qwen |
| Weights | No public weights listed | No public weights listed |
| Released | January 14, 2026 | April 2, 2026 |
| Context window | 400K | 1M |
| Input $/Mtok | $1.75 | $0.325 |
| Output $/Mtok | $14.00 | $1.95 |
| Coding index | 45.4 | 44.1 |
| Independent measurements behind it | 3 | 3 |
| Cost per solved task | $0.43 | $1.19 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is GPT-5.2-Codex better than Qwen3.6 Plus for coding?
GPT-5.2-Codex and Qwen3.6 Plus are level for coding, 45.4 against 44.1 on the Vibe Coding Index, so price and context decide. Pick Qwen3.6 Plus for output tokens 7.2x cheaper ($1.95 vs $14.00 per million) and a larger context window (1M against 400K tokens). Pick GPT-5.2-Codex for a lower cost per solved task ($0.43 against $1.19) and work measured by LiveCodeBench (Vals AI) or Vibe Code Bench (Vals AI), where it leads.
Which is cheaper, GPT-5.2-Codex or Qwen3.6 Plus?
GPT-5.2-Codex costs $1.75 per million input tokens and $14.00 per million output tokens; Qwen3.6 Plus costs $0.325 and $1.95. Per solved SWE-bench Verified task, GPT-5.2-Codex costs $0.43 and Qwen3.6 Plus $1.19.
Which has the bigger context window, GPT-5.2-Codex or Qwen3.6 Plus?
Qwen3.6 Plus, with 1M tokens against 400K.
How many benchmarks is the GPT-5.2-Codex vs Qwen3.6 Plus comparison based on?
14 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 Qwen3.6 Plus's on 3. Where only one model has a result, the page says so instead of naming a winner.