GPT-4 vs Qwen2.5 Coder 32B Instruct
GPT-4 and Qwen2.5 Coder 32B Instruct tie on the Artificial Analysis Intelligence Index at 6.7, and Qwen2.5 Coder 32B Instruct costs 60.0x less per output token ($1.00 vs $60.00 per million). Updated October 21, 2026; 32 benchmarks have a result for both.
GPT-4 or Qwen2.5 Coder 32B Instruct: which to pick
Neither model has a coding score yet, so the two cannot be ranked on coding.
Pick GPT-4 for work measured by SWE-bench Verified or LMArena Text: coding, where it leads.
- Shared results
- 32
- GPT-4 only
- 13
- Qwen2.5 Coder 32B Instruct only
- 3
- Like-for-like categories
- 6 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
Neither has a coding score yet.
One side not measured
Intelligence
AA Intelligence Index
No call
Tied at 6.7.
Same board, both measured
Long documents
Listed context window
Qwen2.5 Coder 32B Instruct
Qwen2.5 Coder 32B Instruct takes 33K tokens against 8K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Qwen2.5 Coder 32B Instruct
GPT-4 $0.060, Qwen2.5 Coder 32B Instruct $0.0012 per turn.
List prices
At a glance
- Coding index
- GPT-4 not benchmarked, Qwen2.5 Coder 32B Instruct not benchmarked.
- Widest category
- Coding, 15.6 points apart on average over 3 shared benchmarks.
- Widest single gap
- AI2 Reasoning Challenge: GPT-4 96.3% against Qwen2.5 Coder 32B Instruct 70.5%, 25.8 points apart on a 0 to 100 scale.
- Output price
- GPT-4 $60.00 against Qwen2.5 Coder 32B Instruct $1.00 per million output tokens, 60.0x apart.
- Context window
- GPT-4 takes 8K 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-4 | Bars | Qwen2.5 Coder 32B Instruct | Ties |
|---|---|---|---|---|
| Coding | 0 | 3 | ||
| Intelligence | 1 | 0 | 1 tie | |
| Reasoning | 0 | 4 | ||
| Math | 0 | 3 | ||
| Knowledge | 2 | 5 | ||
| General | 5 | 8 | ||
| All shared boards | 8 | 23 | 1 tie |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index03
| Benchmark | GPT-4 | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| Aider polyglot | Not measured | 16.4% | Only Qwen2.5 Coder 32B Instruct measured |
| SWE-bench Verified | 22.4% | 47.0% | Qwen2.5 Coder 32B Instruct |
| SWE-bench Lite | 39.7% | Not measured | Only GPT-4 measured |
| SWE-bench (full test set) | 12.5% | Not measured | Only GPT-4 measured |
| AA Coding Index | 13.1 | Not measured | Only GPT-4 measured |
| LiveCodeBench (Artificial Analysis run) | Not measured | 29.5% | Only Qwen2.5 Coder 32B Instruct measured |
| LMArena Text: coding | 1209 | 1276 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: software and IT | 1195 | 1261 | Qwen2.5 Coder 32B Instruct |
| Neither model has a result on the other 31 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | GPT-4 | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| AA Intelligence Index | 6.7 | 6.7 | Tie |
| Epoch Capabilities Index | 125.7 | 119.5 | GPT-4 |
Agentic workNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | GPT-4 | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| Neither model has a result on 41 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index04
| Benchmark | GPT-4 | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| GPQA Diamond (Artificial Analysis run) | 34.9% | 41.7% | Qwen2.5 Coder 32B Instruct |
| Humanity's Last Exam (Artificial Analysis run) | Not measured | 3.5% | Only Qwen2.5 Coder 32B Instruct measured |
| BIG-Bench Hard | 75.1% | Not measured | Only GPT-4 measured |
| LMArena Text: hard prompts | 1200 | 1251 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: expert | 1149 | 1221 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: hard prompts in English | 1209 | 1269 | Qwen2.5 Coder 32B Instruct |
| Neither model has a result on the other 16 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index03
| Benchmark | GPT-4 | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| GSM8K | 92.0% | 93.0% | Qwen2.5 Coder 32B Instruct |
| LMArena Text: math | 1230 | 1251 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: mathematical work | 1228 | 1243 | Qwen2.5 Coder 32B Instruct |
| Neither model has a result on the other 10 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index25
| Benchmark | GPT-4 | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| MMLU Pro (Artificial Analysis run) | 56.2% | 63.5% | Qwen2.5 Coder 32B Instruct |
| MMLU | 86.4% | 79.1% | GPT-4 |
| AI2 Reasoning Challenge | 96.3% | 70.5% | GPT-4 |
| LMArena Text: science | 1177 | 1221 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: medicine and healthcare | 1145 | 1187 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: legal and government | 1191 | 1257 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: business and finance | 1145 | 1224 | Qwen2.5 Coder 32B Instruct |
| 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-4 | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| Neither model has a result on 9 boards here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index58
| Benchmark | GPT-4 | Qwen2.5 Coder 32B Instruct | Leader |
|---|---|---|---|
| IFBench | 33.2% | Not measured | Only GPT-4 measured |
| DTBench | 62.7% | Not measured | Only GPT-4 measured |
| LMCA | 17.1% | Not measured | Only GPT-4 measured |
| HellaSwag | 95.3% | 83.0% | GPT-4 |
| WinoGrande | 87.5% | 80.8% | GPT-4 |
| LMArena Text | 1206 | 1230 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: instruction following | 1201 | 1223 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: creative writing | 1191 | 1174 | GPT-4 |
| LMArena Text: English | 1220 | 1248 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: non-English | 1194 | 1205 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: Chinese | 1184 | 1222 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: German | 1198 | Not measured | Only GPT-4 measured |
| LMArena Text: Japanese | 1137 | Not measured | Only GPT-4 measured |
| LMArena Text: Korean | 1087 | Not measured | Only GPT-4 measured |
| LMArena Text: French | 1220 | Not measured | Only GPT-4 measured |
| LMArena Text: Spanish | 1197 | Not measured | Only GPT-4 measured |
| LMArena Text: Russian | 1195 | 1228 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: Polish | 1265 | Not measured | Only GPT-4 measured |
| LMArena Text: multi-turn | 1206 | 1222 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: longer queries | 1190 | 1251 | Qwen2.5 Coder 32B Instruct |
| LMArena Text: writing and language | 1219 | 1191 | GPT-4 |
| LMArena Text: entertainment and media | 1184 | 1150 | GPT-4 |
| Neither model has a result on the other 24 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | GPT-4 | 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-4 | Qwen2.5 Coder 32B Instruct |
|---|---|---|
| Maker | OpenAI | Qwen |
| Weights | No public weights listed | Open weights |
| Released | May 28, 2023 | November 11, 2024 |
| Context window | 8K | 33K |
| Input $/Mtok | $30.00 | $0.66 |
| Output $/Mtok | $60.00 | $1.00 |
| Coding index | Not benchmarked | Not benchmarked |
| Independent measurements behind it | 0 | 0 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is GPT-4 better than Qwen2.5 Coder 32B Instruct for coding?
Neither model has a coding score yet, so the two cannot be ranked on coding. Pick GPT-4 for work measured by SWE-bench Verified or LMArena Text: coding, where it leads.
Which is cheaper, GPT-4 or Qwen2.5 Coder 32B Instruct?
GPT-4 costs $30.00 per million input tokens and $60.00 per million output tokens; Qwen2.5 Coder 32B Instruct costs $0.66 and $1.00. Cost per solved task is known for neither of them.
Which has the bigger context window, GPT-4 or Qwen2.5 Coder 32B Instruct?
Qwen2.5 Coder 32B Instruct, with 33K tokens against 8K.
How many benchmarks is the GPT-4 vs Qwen2.5 Coder 32B Instruct comparison based on?
32 benchmarks have a result for both models as of October 21, 2026. GPT-4's coding index rests on 0 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.