GPT-5.2-Codex vs Qwen2.5 Coder 32B Instruct
GPT-5.2-Codex leads Qwen2.5 Coder 32B Instruct on the Artificial Analysis Intelligence Index, 28.5 to 6.7, and Qwen2.5 Coder 32B Instruct costs 14.0x less per output token ($1.00 vs $14.00 per million). Updated October 22, 2026; 4 benchmarks have a result for both.
GPT-5.2-Codex or Qwen2.5 Coder 32B Instruct: which to pick
Only GPT-5.2-Codex has a coding score, 45.4 on the Vibe Coding Index; Qwen2.5 Coder 32B Instruct has none, so the two cannot be ranked on coding.
Pick Qwen2.5 Coder 32B Instruct for output tokens 14.0x cheaper ($1.00 vs $14.00 per million).
- Shared results
- 4
- GPT-5.2-Codex only
- 18
- Qwen2.5 Coder 32B Instruct only
- 31
- Like-for-like categories
- 3 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-5.2-Codex has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
GPT-5.2-Codex
GPT-5.2-Codex leads, 28.5 against 6.7.
Same board, both measured
Long documents
Listed context window
GPT-5.2-Codex
GPT-5.2-Codex takes 400K tokens against 33K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Qwen2.5 Coder 32B Instruct
GPT-5.2-Codex $0.0088, Qwen2.5 Coder 32B Instruct $0.0012 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only GPT-5.2-Codex has the number, $0.43.
Directional only
At a glance
- Coding index
- GPT-5.2-Codex 45.4 from 3 benchmarks, Qwen2.5 Coder 32B Instruct not benchmarked.
- Widest category
- Reasoning, 40.2 points apart on average over 2 shared benchmarks.
- Widest single gap
- GPQA Diamond (Artificial Analysis run): GPT-5.2-Codex 89.9% against Qwen2.5 Coder 32B Instruct 41.7%, 48.2 points apart on a 0 to 100 scale.
- Output price
- GPT-5.2-Codex $14.00 against Qwen2.5 Coder 32B Instruct $1.00 per million output tokens, 14.0x apart.
- Context window
- GPT-5.2-Codex takes 400K tokens, Qwen2.5 Coder 32B Instruct 33K.
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 | Qwen2.5 Coder 32B Instruct | Ties |
|---|---|---|---|---|
| Coding | 1 | 0 | ||
| Intelligence | 1 | 0 | ||
| Reasoning | 2 | 0 | ||
| All shared boards | 4 | 0 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index10
| Benchmark | GPT-5.2-Codex | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| Aider polyglot | Not measured | 16.4% | Only Qwen2.5 Coder 32B Instruct measured |
| SWE-bench Verified | 72.8% | 47.0% | GPT-5.2-Codex |
| SWE-bench Multilingual | 66.3% | Not measured | Only GPT-5.2-Codex measured |
| LMArena WebDev | 1339 | Not measured | Only GPT-5.2-Codex measured |
| LiveCodeBench (Vals AI) | 88.0% | Not measured | Only GPT-5.2-Codex measured |
| SWE-bench Verified (Vals AI) | 72.4% | Not measured | Only GPT-5.2-Codex measured |
| Vibe Code Bench (Vals AI) | 37.9% | Not measured | Only GPT-5.2-Codex measured |
| LiveCodeBench (Artificial Analysis run) | Not measured | 29.5% | Only Qwen2.5 Coder 32B Instruct measured |
| Terminal-Bench Hard (Artificial Analysis run) | 37.1% | Not measured | Only GPT-5.2-Codex measured |
| Terminal-Bench | 66.5% | Not measured | Only GPT-5.2-Codex measured |
| ALE-Bench | 1299.9 | Not measured | Only GPT-5.2-Codex measured |
| LMArena Text: coding | Not measured | 1276 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: software and IT | Not measured | 1261 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena WebDev: React | 1329 | Not measured | Only GPT-5.2-Codex measured |
| LMArena WebDev: HTML | 1348 | Not measured | Only GPT-5.2-Codex measured |
| Neither model has a result on the other 24 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | GPT-5.2-Codex | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| AA Intelligence Index | 28.5 | 6.7 | GPT-5.2-Codex |
| Epoch Capabilities Index | Not measured | 119.5 | Only Qwen2.5 Coder 32B Instruct measured |
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index6 boards, none shared
| Benchmark | GPT-5.2-Codex | Qwen2.5 Coder 32B Instruct | 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% | Not measured | Only GPT-5.2-Codex measured |
| Neither model has a result on the other 35 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | GPT-5.2-Codex | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| GPQA Diamond (Artificial Analysis run) | 89.9% | 41.7% | GPT-5.2-Codex |
| Humanity's Last Exam (Artificial Analysis run) | 35.7% | 3.5% | GPT-5.2-Codex |
| LMArena Text: hard prompts | Not measured | 1251 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: expert | Not measured | 1221 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: hard prompts in English | Not measured | 1269 | Only Qwen2.5 Coder 32B Instruct measured |
| Neither model has a result on the other 17 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index3 boards, none shared
| Benchmark | GPT-5.2-Codex | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| GSM8K | Not measured | 93.0% | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: math | Not measured | 1251 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: mathematical work | Not measured | 1243 | Only Qwen2.5 Coder 32B Instruct measured |
| Neither model has a result on the other 10 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index7 boards, none shared
| Benchmark | GPT-5.2-Codex | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| MMLU Pro (Artificial Analysis run) | Not measured | 63.5% | Only Qwen2.5 Coder 32B Instruct measured |
| MMLU | Not measured | 79.1% | Only Qwen2.5 Coder 32B Instruct measured |
| AI2 Reasoning Challenge | Not measured | 70.5% | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: science | Not measured | 1221 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: medicine and healthcare | Not measured | 1187 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: legal and government | Not measured | 1257 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: business and finance | Not measured | 1224 | Only Qwen2.5 Coder 32B Instruct measured |
| Neither model has a result on the other 11 boards here | |||
Web designNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | GPT-5.2-Codex | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| Neither model has a result on 9 boards here | |||
GeneralDirectional onlyPublic boards, shown as published, kept out of the index15 boards, none shared
| Benchmark | GPT-5.2-Codex | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| IFBench | 77.6% | Not measured | Only GPT-5.2-Codex measured |
| Long context reasoning (AA-LCR) | 82.3% | Not measured | Only GPT-5.2-Codex measured |
| HellaSwag | Not measured | 83.0% | Only Qwen2.5 Coder 32B Instruct measured |
| WinoGrande | Not measured | 80.8% | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text | Not measured | 1230 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: instruction following | Not measured | 1223 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: creative writing | Not measured | 1174 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: English | Not measured | 1248 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: non-English | Not measured | 1205 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: Chinese | Not measured | 1222 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: Russian | Not measured | 1228 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: multi-turn | Not measured | 1222 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: longer queries | Not measured | 1251 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: writing and language | Not measured | 1191 | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: entertainment and media | Not measured | 1150 | Only Qwen2.5 Coder 32B Instruct measured |
| Neither model has a result on the other 31 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | GPT-5.2-Codex | Qwen2.5 Coder 32B 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 Coder 32B Instruct costs less
Repository review
50K input + 3K output tokens
Qwen2.5 Coder 32B Instruct costs less
Agent loop
200K input + 10K output tokens
Qwen2.5 Coder 32B 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-5.2-Codex | Qwen2.5 Coder 32B Instruct |
|---|---|---|
| Maker | OpenAI | Qwen |
| Weights | No public weights listed | Open weights |
| Released | January 14, 2026 | November 11, 2024 |
| Context window | 400K | 33K |
| Input $/Mtok | $1.75 | $0.66 |
| Output $/Mtok | $14.00 | $1.00 |
| Coding index | 45.4 | Not benchmarked |
| Independent measurements behind it | 3 | 0 |
| Cost per solved task | $0.43 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is GPT-5.2-Codex better than Qwen2.5 Coder 32B Instruct for coding?
Only GPT-5.2-Codex has a coding score, 45.4 on the Vibe Coding Index; Qwen2.5 Coder 32B Instruct has none, so the two cannot be ranked on coding. Pick Qwen2.5 Coder 32B Instruct for output tokens 14.0x cheaper ($1.00 vs $14.00 per million).
Which is cheaper, GPT-5.2-Codex or Qwen2.5 Coder 32B Instruct?
GPT-5.2-Codex costs $1.75 per million input tokens and $14.00 per million output tokens; Qwen2.5 Coder 32B Instruct costs $0.66 and $1.00. Cost per solved task is known for GPT-5.2-Codex only.
Which has the bigger context window, GPT-5.2-Codex or Qwen2.5 Coder 32B Instruct?
GPT-5.2-Codex, with 400K tokens against 33K.
How many benchmarks is the GPT-5.2-Codex vs Qwen2.5 Coder 32B Instruct comparison based on?
4 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 Qwen2.5 Coder 32B Instruct's on 0. Where only one model has a result, the page says so instead of naming a winner.