# GPT-5.6 Luna vs Qwen3 32B

> GPT-5.6 Luna is the stronger model overall, scoring 54.6 to 39.2 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gpt-5-6-luna-vs-qwen3-32b
- Last updated: 2026-10-10
- Shared benchmarks: 22

## Summary

- They share 22 benchmarks with published results for both. GPT-5.6 Luna scores higher in 9 categories and Qwen3 32B in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.6 Luna leads 77.7 to 39.7.
- The biggest single-benchmark swing is Chess Puzzles: 40% for GPT-5.6 Luna and 5% for Qwen3 32B.
- GPT-5.6 Luna is cheaper at $0.20 / $1.20 per million input/output tokens, against $0.70 / $2.80 for Qwen3 32B.
- GPT-5.6 Luna accepts more context: 1.05M tokens versus 131K.
- Qwen3 32B has downloadable open weights; the other is API-only.

## Snapshot

| | GPT-5.6 Luna | Qwen3 32B |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 54.6 | 39.2 |
| Rank | 30 | 172 |
| Context | 1.05M | 131K |
| Input $/M | $0.20 | $0.70 |
| Output $/M | $1.20 | $2.80 |
| Weights | Proprietary | Open |

## Coding

- GPT-5.6 Luna: 54.5 (#28)
- Qwen3 32B: 37.7 (#190)

| Benchmark | GPT-5.6 Luna | Qwen3 32B |
|---|---|---|
| SciCode | 53.6% | 35.4% |
| LMArena Coding | 1466 | 1358 |
| DeepSWE | 67.2% | — |
| FrontierCode | 39.8% | — |
| Aider Polyglot | — | 40% |
| CursorBench | 35.9% | — |
| LMArena WebDev | 1519 | — |
| WeirdML | 60.9% | — |
| ALE-Bench | 1,667 | — |

## Agentic & Tool Use

- GPT-5.6 Luna: 34.4 (#45)
- Qwen3 32B: 32.6 (#62)

| Benchmark | GPT-5.6 Luna | Qwen3 32B |
|---|---|---|
| APEX-Agents | 43% | — |
| Berkeley Function Calling Leaderboard | — | 48.7% |
| BALROG | 45.6% | — |
| GDP.pdf | 22.7% | — |
| Vending-Bench 2 | 4,095 | — |

## Reasoning

- GPT-5.6 Luna: 47.6 (#43)
- Qwen3 32B: 20.2 (#241)

| Benchmark | GPT-5.6 Luna | Qwen3 32B |
|---|---|---|
| Kagi LLM Benchmark | 49.1% | 54.9% |
| CritPt | 20.6% | 0.3% |
| Chess Puzzles | 40% | 5% |
| LMArena Hard Prompts | 1451 | 1334 |
| DTBench | 89.1% | 67.5% |
| LMCA | 48.5% | 17.3% |
| Epoch Capabilities Index | 156.39 | 138.51 |
| ARC-AGI-2 | 59.5% | — |
| SimpleBench | 46.8% | — |
| NYT Connections (extended) | 69.4% | — |
| ARC-AGI-1 | 88% | — |
| Mystery Game Puzzles | 21% | — |
| Surface Evolver Bench | 61.9% | — |

## Math

- GPT-5.6 Luna: 77.7 (#14)
- Qwen3 32B: 39.7 (#99)

| Benchmark | GPT-5.6 Luna | Qwen3 32B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 98.3% | 66.9% |
| LMArena Math | 1458 | 1399 |
| FrontierMath (Tiers 1-3) | 82.1% | — |
| FrontierMath Tier 4 | 61% | — |
| ProofBench | 60% | — |

## Knowledge

- GPT-5.6 Luna: 58.5 (#34)
- Qwen3 32B: 40.0 (#125)

| Benchmark | GPT-5.6 Luna | Qwen3 32B |
|---|---|---|
| GPQA Diamond | 91.6% | 65.7% |
| LMArena Expert | 1478 | 1362 |
| SimpleQA Verified | 41% | — |
| Vectara Hallucination Rate | — | 5.9% |

## Multimodal

- GPT-5.6 Luna: 42.7 (#28)
- Qwen3 32B: —

| Benchmark | GPT-5.6 Luna | Qwen3 32B |
|---|---|---|
| LMArena Vision | 1258 | — |
| Blueprint-Bench 2 | 22.6% | — |
| Furniture Assembly | 42.5% | — |
| LMArena Document | 1457 | — |

## Multilingual

- GPT-5.6 Luna: 52.8 (#78)
- Qwen3 32B: 45.6 (#167)

| Benchmark | GPT-5.6 Luna | Qwen3 32B |
|---|---|---|
| LMArena Non-English | 1417 | 1317 |
| LMArena Chinese | 1470 | 1357 |
| LMArena German | 1454 | 1341 |
| LMArena Russian | 1428 | 1311 |
| LMArena French | 1456 | — |
| LMArena Japanese | 1411 | — |
| LMArena Korean | 1415 | — |
| LMArena Spanish | 1448 | — |

## Instruction Following

- GPT-5.6 Luna: 75.6 (#57)
- Qwen3 32B: 68.9 (#179)

| Benchmark | GPT-5.6 Luna | Qwen3 32B |
|---|---|---|
| LMArena Instruction Following | 1437 | 1305 |

## Long Context

- GPT-5.6 Luna: 43.9 (#82)
- Qwen3 32B: 43.8 (#87)

| Benchmark | GPT-5.6 Luna | Qwen3 32B |
|---|---|---|
| LMArena Longer Query | 1436 | 1327 |
| Fiction.LiveBench | — | 74.2% |

## Writing & Preference

- GPT-5.6 Luna: 68.0 (#29)
- Qwen3 32B: 52.9 (#163)

| Benchmark | GPT-5.6 Luna | Qwen3 32B |
|---|---|---|
| LMArena Text | 1431 | 1340 |
| LMArena Creative Writing | 1396 | 1297 |
| LMArena Multi-Turn | 1434 | 1331 |
| EQ-Bench Creative Writing | 1829 | — |
| EQ-Bench 4 | 1156 | — |

## FAQ

### Is GPT-5.6 Luna better than Qwen3 32B?

GPT-5.6 Luna is the stronger model overall, scoring 54.6 to 39.2 on the Noometry Index.

### Which is cheaper, GPT-5.6 Luna or Qwen3 32B?

GPT-5.6 Luna is cheaper. It lists at $0.20 per million input tokens and $1.20 per million output tokens; Qwen3 32B lists at $0.70 and $2.80.

### Is GPT-5.6 Luna or Qwen3 32B better for coding?

GPT-5.6 Luna scores higher on coding benchmarks: 54.5 versus 37.7 in the Noometry coding category.

### Which has the bigger context window?

GPT-5.6 Luna does, with 1.05M tokens against 131K.

### How many benchmarks do GPT-5.6 Luna and Qwen3 32B share?

22 benchmarks have published results for both models. GPT-5.6 Luna has 52 scored results on Noometry and Qwen3 32B has 26.