Model comparison
GPT-4o mini vs Qwen3 235B-A22B
Qwen3 235B-A22B is the stronger model overall, scoring 43.2 to 25.4 on the Noometry Index. GPT-4o mini costs 4.7× less per token, which makes it the better buy when Qwen3 235B-A22B's lead doesn't matter for your workload.
Last verified . 39 shared benchmarks.
Alibaba (Qwen)Summary
- They share 39 benchmarks with published results for both. GPT-4o mini scores higher in 0 categories and Qwen3 235B-A22B in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen3 235B-A22B leads 48.5 to 9.9.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 6.9% for GPT-4o mini and 86.7% for Qwen3 235B-A22B.
- GPT-4o mini is cheaper at $0.15 / $0.60 per million input/output tokens, against $0.70 / $2.80 for Qwen3 235B-A22B.
- Qwen3 235B-A22B accepts more context: 131K tokens versus 128K.
- Qwen3 235B-A22B has downloadable open weights; the other is API-only.
Side by side
| GPT-4o mini | Qwen3 235B-A22B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 25.4 | 43.2 |
| Released | 2024-07-18 | 2025-04 |
| Weights | Proprietary | Open |
| Context window | 128K | 131K |
| Max output | 16K | 16K |
| Input $ / M tokens | $0.15 | $0.70 |
| Output $ / M tokens | $0.60 | $2.80 |
| Results tracked | 60 | 49 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3 235B-A22B leads
GPT-4o mini: 21.6 (#330), Qwen3 235B-A22B: 43.8 (#72)
| Benchmark | GPT-4o mini | Qwen3 235B-A22B |
|---|---|---|
| Aider Polyglot | 3.6% | 59.6% |
| WeirdML | 11.8% | 41% |
| LMArena Coding | 1290 | 1445 |
| SciCode | — | 42.4% |
| BigCodeBench Instruct | 46.1% | — |
| LiveBench Coding | 43.1% | — |
| BigCodeBench Complete | 57.4% | — |
| HumanEval+ | 83.5% | — |
| MBPP+ | 72.2% | — |
Agentic & Tool Use Qwen3 235B-A22B leads
GPT-4o mini: 27.2 (#95), Qwen3 235B-A22B: 33.7 (#48)
| Benchmark | GPT-4o mini | Qwen3 235B-A22B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 52.1% |
| BALROG | 17.4% | — |
| Vending-Bench 2 | — | -11.34 |
Reasoning Qwen3 235B-A22B leads
GPT-4o mini: 8.7 (#343), Qwen3 235B-A22B: 15.8 (#302)
| Benchmark | GPT-4o mini | Qwen3 235B-A22B |
|---|---|---|
| ARC-AGI-2 | 0% | 1.3% |
| SimpleBench | 10.7% | 31% |
| Kagi LLM Benchmark | 28.8% | 69.4% |
| Chess Puzzles | 0% | 12% |
| LMArena Hard Prompts | 1267 | 1433 |
| Mystery Game Puzzles | 12% | 9% |
| DTBench | 54.4% | 80.3% |
| LMCA | 10.4% | 29.3% |
| Epoch Capabilities Index | 126.56 | 143.87 |
| ARC-AGI-1 | — | 11% |
| CritPt | — | 0% |
| LiveBench Reasoning | 32.8% | — |
| LiveBench Data Analysis | 50% | — |
| ForecastBench | — | 59.7 |
| LiveBench | 41.3% | — |
| PIQA | 88.7% | — |
Math Qwen3 235B-A22B leads
GPT-4o mini: 9.9 (#307), Qwen3 235B-A22B: 48.5 (#51)
| Benchmark | GPT-4o mini | Qwen3 235B-A22B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 6.9% | 86.7% |
| Omni-MATH | 28% | 71.8% |
| LMArena Math | 1267 | 1431 |
| MATH Level 5 | 52.6% | 68.9% |
| FrontierMath (Tiers 1-3) | 0.7% | — |
| LiveBench Math | 36.3% | — |
| FrontierMath (Feb 2025 set) | — | 8.5% |
| FrontierMath Tier 4 (v1) | — | 0% |
| GSM8K | 91.3% | — |
Knowledge Qwen3 235B-A22B leads
GPT-4o mini: 17.7 (#279), Qwen3 235B-A22B: 49.5 (#67)
| Benchmark | GPT-4o mini | Qwen3 235B-A22B |
|---|---|---|
| GPQA Diamond | 37.7% | 80.1% |
| SimpleQA Verified | 8.3% | 40.4% |
| MMLU-Pro | 60.3% | 84.4% |
| Confabulations | 37.2% | 15.6% |
| GPQA (HELM) | 36.8% | 72.7% |
| LMArena Expert | 1235 | 1460 |
| Vectara Hallucination Rate | — | 9.3% |
| BoolQ | 88.7% | — |
| MMLU | 81.8% | — |
Multimodal Not comparable
GPT-4o mini: 25.3 (#117), Qwen3 235B-A22B: —
| Benchmark | GPT-4o mini | Qwen3 235B-A22B |
|---|---|---|
| LMArena Vision | 1066 | — |
| Video-MME | 64.8% | — |
| GeoBench | 64% | — |
| VPCT | 34% | — |
Multilingual Qwen3 235B-A22B leads
GPT-4o mini: 42.2 (#194), Qwen3 235B-A22B: 52.5 (#84)
| Benchmark | GPT-4o mini | Qwen3 235B-A22B |
|---|---|---|
| LMArena Non-English | 1266 | 1410 |
| LMArena Chinese | 1264 | 1481 |
| LMArena French | 1296 | 1445 |
| LMArena German | 1272 | 1432 |
| LMArena Japanese | 1216 | 1400 |
| LMArena Korean | 1195 | 1391 |
| LMArena Russian | 1275 | 1411 |
| LMArena Spanish | 1276 | 1429 |
Instruction Following Qwen3 235B-A22B leads
GPT-4o mini: 62.1 (#234), Qwen3 235B-A22B: 72.8 (#132)
| Benchmark | GPT-4o mini | Qwen3 235B-A22B |
|---|---|---|
| IFEval | 78.2% | 83.5% |
| LMArena Instruction Following | 1259 | 1408 |
| LiveBench Instruction Following | 56.8% | — |
Long Context Qwen3 235B-A22B leads
GPT-4o mini: 39.1 (#181), Qwen3 235B-A22B: 46.1 (#25)
| Benchmark | GPT-4o mini | Qwen3 235B-A22B |
|---|---|---|
| LMArena Longer Query | 1289 | 1426 |
| Fiction.LiveBench | — | 75% |
Writing & Preference Qwen3 235B-A22B leads
GPT-4o mini: 39.7 (#243), Qwen3 235B-A22B: 59.8 (#103)
| Benchmark | GPT-4o mini | Qwen3 235B-A22B |
|---|---|---|
| LMArena Text | 1286 | 1419 |
| LMArena Creative Writing | 1268 | 1384 |
| Short-Story Creative Writing | 67.2% | 83% |
| EQ-Bench Creative Writing | 873 | 1366 |
| WildBench | 79.1% | 86.6% |
| LMArena Multi-Turn | 1285 | 1432 |
| LiveBench Language | 28.6% | — |
Frequently asked questions
Is GPT-4o mini better than Qwen3 235B-A22B?
Qwen3 235B-A22B is the stronger model overall, scoring 43.2 to 25.4 on the Noometry Index. GPT-4o mini costs 4.7× less per token, which makes it the better buy when Qwen3 235B-A22B's lead doesn't matter for your workload.
Which is cheaper, GPT-4o mini or Qwen3 235B-A22B?
GPT-4o mini is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; Qwen3 235B-A22B lists at $0.70 and $2.80.
Is GPT-4o mini or Qwen3 235B-A22B better for coding?
Qwen3 235B-A22B scores higher on coding benchmarks: 43.8 versus 21.6 in the Noometry coding category.
Which has the bigger context window?
Qwen3 235B-A22B does, with 131K tokens against 128K.
How many benchmarks do GPT-4o mini and Qwen3 235B-A22B share?
39 benchmarks have published results for both models. GPT-4o mini has 60 scored results on Noometry and Qwen3 235B-A22B has 49.