Model comparison
GPT-6 Luna vs Qwen2.5 32B Instruct
GPT-6 Luna is the stronger model overall, scoring 45.1 to 29.7 on the Noometry Index.
Last verified . 0 shared benchmarks.
Alibaba (Qwen)Summary
- The widest gap is in math, where GPT-6 Luna leads 35.6 to 15.3.
- GPT-6 Luna is cheaper at $0.10 / $0.50 per million input/output tokens, against $0.70 / $2.80 for Qwen2.5 32B Instruct.
- GPT-6 Luna accepts more context: 1.05M tokens versus 131K.
- Qwen2.5 32B Instruct has downloadable open weights; the other is API-only.
Side by side
| GPT-6 Luna | Qwen2.5 32B Instruct | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 45.1 | 29.7 |
| Released | 2026-09-22 | 2024-09 |
| Weights | Proprietary | Open |
| Context window | 1.05M | 131K |
| Max output | 128K | 8K |
| Input $ / M tokens | $0.10 | $0.70 |
| Output $ / M tokens | $0.50 | $2.80 |
| Results tracked | 18 | 7 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Luna leads
GPT-6 Luna: 54.7 (#24), Qwen2.5 32B Instruct: 38.1 (#165)
| Benchmark | GPT-6 Luna | Qwen2.5 32B Instruct |
|---|---|---|
| DeepSWE | 66.6% | — |
| FrontierCode | 42.4% | — |
| LMArena WebDev | 1583 | — |
| BigCodeBench Instruct | — | 45% |
| LMArena Coding | 1448 | — |
| BigCodeBench Complete | — | 52.3% |
Reasoning GPT-6 Luna leads
GPT-6 Luna: 31.7 (#68), Qwen2.5 32B Instruct: 18.5 (#262)
| Benchmark | GPT-6 Luna | Qwen2.5 32B Instruct |
|---|---|---|
| NYT Connections (extended) | 68.7% | — |
| Chess Puzzles | — | 0% |
| LMArena Hard Prompts | 1415 | — |
| Epoch Capabilities Index | — | 128.52 |
Math GPT-6 Luna leads
GPT-6 Luna: 35.6 (#140), Qwen2.5 32B Instruct: 15.3 (#289)
| Benchmark | GPT-6 Luna | Qwen2.5 32B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 7.4% |
| LMArena Math | 1396 | — |
| MATH Level 5 | — | 56.1% |
Knowledge GPT-6 Luna leads
GPT-6 Luna: 40.1 (#110), Qwen2.5 32B Instruct: 24.9 (#259)
| Benchmark | GPT-6 Luna | Qwen2.5 32B Instruct |
|---|---|---|
| GPQA Diamond | — | 46.1% |
| LMArena Expert | 1458 | — |
Multimodal Not comparable
GPT-6 Luna: 42.0 (#27), Qwen2.5 32B Instruct: —
| Benchmark | GPT-6 Luna | Qwen2.5 32B Instruct |
|---|---|---|
| Furniture Assembly | 44.2% | — |
Multilingual Not comparable
GPT-6 Luna: 51.0 (#111), Qwen2.5 32B Instruct: —
| Benchmark | GPT-6 Luna | Qwen2.5 32B Instruct |
|---|---|---|
| LMArena Non-English | 1390 | — |
| LMArena Chinese | 1425 | — |
| LMArena Russian | 1392 | — |
| LMArena Spanish | 1416 | — |
Instruction Following Not comparable
GPT-6 Luna: 74.7 (#85), Qwen2.5 32B Instruct: —
| Benchmark | GPT-6 Luna | Qwen2.5 32B Instruct |
|---|---|---|
| LMArena Instruction Following | 1414 | — |
Long Context Not comparable
GPT-6 Luna: 43.3 (#95), Qwen2.5 32B Instruct: —
| Benchmark | GPT-6 Luna | Qwen2.5 32B Instruct |
|---|---|---|
| LMArena Longer Query | 1417 | — |
Writing & Preference Not comparable
GPT-6 Luna: 58.4 (#115), Qwen2.5 32B Instruct: —
| Benchmark | GPT-6 Luna | Qwen2.5 32B Instruct |
|---|---|---|
| LMArena Text | 1392 | — |
| LMArena Creative Writing | 1357 | — |
| LMArena Multi-Turn | 1400 | — |
Frequently asked questions
Is GPT-6 Luna better than Qwen2.5 32B Instruct?
GPT-6 Luna is the stronger model overall, scoring 45.1 to 29.7 on the Noometry Index.
Which is cheaper, GPT-6 Luna or Qwen2.5 32B Instruct?
GPT-6 Luna is cheaper. It lists at $0.10 per million input tokens and $0.50 per million output tokens; Qwen2.5 32B Instruct lists at $0.70 and $2.80.
Is GPT-6 Luna or Qwen2.5 32B Instruct better for coding?
GPT-6 Luna scores higher on coding benchmarks: 54.7 versus 38.1 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Luna does, with 1.05M tokens against 131K.
How many benchmarks do GPT-6 Luna and Qwen2.5 32B Instruct share?
0 benchmarks have published results for both models. GPT-6 Luna has 18 scored results on Noometry and Qwen2.5 32B Instruct has 7.