Model comparison
Gemma 4 31B IT vs GPT-6 Astra
GPT-6 Astra is the stronger model overall, scoring 70.8 to 43.5 on the Noometry Index. Gemma 4 31B IT costs 131× less per token, which makes it the better buy when GPT-6 Astra's lead doesn't matter for your workload.
Last verified . 31 shared benchmarks.
GoogleSummary
- They share 31 benchmarks with published results for both. Gemma 4 31B IT scores higher in 1 category and GPT-6 Astra in 8 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-6 Astra leads 85.1 to 27.2.
- The biggest single-benchmark swing is Chess Puzzles: 5% for Gemma 4 31B IT and 72% for GPT-6 Astra.
- Gemma 4 31B IT is cheaper at $0.09 / $0.34 per million input/output tokens, against $10 / $50 for GPT-6 Astra.
- GPT-6 Astra accepts more context: 1.05M tokens versus 262K.
- Gemma 4 31B IT has downloadable open weights; the other is API-only.
Side by side
| Gemma 4 31B IT | GPT-6 Astra | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 43.5 | 70.8 |
| Released | 2026-04-02 | 2026-09-03 |
| Weights | Open | Proprietary |
| Context window | 262K | 1.05M |
| Max output | 33K | 128K |
| Input $ / M tokens | $0.09 | $10 |
| Output $ / M tokens | $0.34 | $50 |
| Results tracked | 35 | 56 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Astra leads
Gemma 4 31B IT: 42.3 (#108), GPT-6 Astra: 73.7 (#2)
| Benchmark | Gemma 4 31B IT | GPT-6 Astra |
|---|---|---|
| LMArena WebDev | 1366 | 1786 |
| SciCode | 43.4% | 56.5% |
| WeirdML | 52.3% | 93.6% |
| LMArena Coding | 1459 | 1487 |
| ALE-Bench | 925.5 | 2,951 |
| DeepSWE | — | 74.1% |
| FrontierCode | — | 53.3% |
| FrontierSWE | — | 65.5% |
| GSO | — | 79.4% |
| MirrorCode | — | 46.7% |
Agentic & Tool Use Not comparable
Gemma 4 31B IT: —, GPT-6 Astra: 52.9 (#3)
| Benchmark | Gemma 4 31B IT | GPT-6 Astra |
|---|---|---|
| APEX-Agents | — | 64.7% |
| Remote Labor Index | — | 20.8% |
| BALROG | — | 68.3% |
| GDP.pdf | — | 34.2% |
| Vending-Bench 2 | — | 15,515 |
Reasoning GPT-6 Astra leads
Gemma 4 31B IT: 27.2 (#122), GPT-6 Astra: 85.1 (#1)
| Benchmark | Gemma 4 31B IT | GPT-6 Astra |
|---|---|---|
| NYT Connections (extended) | 70.6% | 98.1% |
| CritPt | 1.4% | 31.7% |
| Chess Puzzles | 5% | 72% |
| LMArena Hard Prompts | 1448 | 1462 |
| DTBench | 82.7% | 97.3% |
| LMCA | 39.3% | 64.4% |
| Epoch Capabilities Index | 142.74 | 166.45 |
| ARC-AGI-2 | — | 95% |
| Kagi LLM Benchmark | 63.5% | — |
| ARC-AGI-1 | — | 98.5% |
| Thematic Generalization | 53% | — |
| EBR-Bench | — | 76.2% |
| Mystery Game Puzzles | — | 84% |
| Surface Evolver Bench | 30.6% | — |
| Bench to the Future 3 | — | 0.14 |
Math GPT-6 Astra leads
Gemma 4 31B IT: 43.2 (#81), GPT-6 Astra: 93.5 (#2)
| Benchmark | Gemma 4 31B IT | GPT-6 Astra |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 73.3% | 100% |
| LMArena Math | 1465 | 1465 |
| FrontierMath (Tiers 1-3) | — | 93.7% |
| FrontierMath Tier 4 | — | 97.6% |
| ProofBench | — | 99% |
| FrontierMath Erdős | — | 2.9% |
Knowledge GPT-6 Astra leads
Gemma 4 31B IT: 37.9 (#151), GPT-6 Astra: 75.3 (#1)
| Benchmark | Gemma 4 31B IT | GPT-6 Astra |
|---|---|---|
| GPQA Diamond | 75.8% | 95.8% |
| SimpleQA Verified | 10.4% | 75.6% |
| Vectara Hallucination Rate | 7.4% | 8.7% |
| LMArena Expert | 1465 | 1483 |
| Humanity's Last Exam | — | 54.8% |
Multimodal GPT-6 Astra leads
Gemma 4 31B IT: 41.6 (#34), GPT-6 Astra: 55.0 (#3)
| Benchmark | Gemma 4 31B IT | GPT-6 Astra |
|---|---|---|
| LMArena Vision | 1277 | 1281 |
| LMArena Document | 1425 | 1468 |
| Blueprint-Bench 2 | — | 49.7% |
| Furniture Assembly | — | 80% |
Multilingual Too close to call
Gemma 4 31B IT: 53.8 (#57), GPT-6 Astra: 53.7 (#61)
| Benchmark | Gemma 4 31B IT | GPT-6 Astra |
|---|---|---|
| LMArena Non-English | 1431 | 1430 |
| LMArena Chinese | 1476 | 1484 |
| LMArena French | 1435 | 1456 |
| LMArena Russian | 1460 | 1436 |
| LMArena Spanish | 1444 | 1407 |
| LMArena German | — | 1440 |
| LMArena Japanese | — | 1379 |
| LMArena Korean | — | 1426 |
Instruction Following Too close to call
Gemma 4 31B IT: 75.5 (#61), GPT-6 Astra: 76.3 (#44)
| Benchmark | Gemma 4 31B IT | GPT-6 Astra |
|---|---|---|
| LMArena Instruction Following | 1433 | 1450 |
Long Context Too close to call
Gemma 4 31B IT: 44.2 (#71), GPT-6 Astra: 44.5 (#62)
| Benchmark | Gemma 4 31B IT | GPT-6 Astra |
|---|---|---|
| LMArena Longer Query | 1446 | 1456 |
Writing & Preference GPT-6 Astra leads
Gemma 4 31B IT: 60.5 (#96), GPT-6 Astra: 75.3 (#7)
| Benchmark | Gemma 4 31B IT | GPT-6 Astra |
|---|---|---|
| LMArena Text | 1443 | 1441 |
| LMArena Creative Writing | 1415 | 1418 |
| EQ-Bench Creative Writing | 1368 | 2173 |
| LMArena Multi-Turn | 1452 | 1448 |
| EQ-Bench 4 | 1120 | — |
Frequently asked questions
Is Gemma 4 31B IT better than GPT-6 Astra?
GPT-6 Astra is the stronger model overall, scoring 70.8 to 43.5 on the Noometry Index. Gemma 4 31B IT costs 131× less per token, which makes it the better buy when GPT-6 Astra's lead doesn't matter for your workload.
Which is cheaper, Gemma 4 31B IT or GPT-6 Astra?
Gemma 4 31B IT is cheaper. It lists at $0.09 per million input tokens and $0.34 per million output tokens; GPT-6 Astra lists at $10 and $50.
Is Gemma 4 31B IT or GPT-6 Astra better for coding?
GPT-6 Astra scores higher on coding benchmarks: 73.7 versus 42.3 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Astra does, with 1.05M tokens against 262K.
How many benchmarks do Gemma 4 31B IT and GPT-6 Astra share?
31 benchmarks have published results for both models. Gemma 4 31B IT has 35 scored results on Noometry and GPT-6 Astra has 56.