Model comparison
Claude Sonnet 4.6 vs Gemma 4 31B IT
Claude Sonnet 4.6 is the stronger model overall, scoring 50.1 to 43.2 on the Noometry Index. Gemma 4 31B IT costs 39× less per token, which makes it the better buy when Claude Sonnet 4.6's lead doesn't matter for your workload.
Last verified . 33 shared benchmarks.
GoogleSummary
- They share 33 benchmarks with published results for both. Claude Sonnet 4.6 scores higher in 8 categories and Gemma 4 31B IT in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Claude Sonnet 4.6 leads 46.1 to 26.4.
- The biggest single-benchmark swing is SimpleQA Verified: 35.5% for Claude Sonnet 4.6 and 10.4% for Gemma 4 31B IT.
- Gemma 4 31B IT is cheaper at $0.09 / $0.34 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.6.
- Claude Sonnet 4.6 accepts more context: 1M tokens versus 262K.
- Gemma 4 31B IT has downloadable open weights; the other is API-only.
Side by side
| Claude Sonnet 4.6 | Gemma 4 31B IT | |
|---|---|---|
| Provider | Anthropic | |
| Noometry Index | 50.1 | 43.2 |
| Released | 2026-02-17 | 2026-04-02 |
| Weights | Proprietary | Open |
| Context window | 1M | 262K |
| Max output | 128K | 33K |
| Input $ / M tokens | $3 | $0.09 |
| Output $ / M tokens | $15 | $0.34 |
| Results tracked | 57 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 46.0 (#62), Gemma 4 31B IT: 41.9 (#98)
| Benchmark | Claude Sonnet 4.6 | Gemma 4 31B IT |
|---|---|---|
| LMArena WebDev | 1521 | 1364 |
| SciCode | 46.8% | 43.4% |
| WeirdML | 66.1% | 52.3% |
| LMArena Coding | 1504 | 1459 |
| ALE-Bench | 1,327 | 925.5 |
| SWE-bench Verified | 75.2% | — |
| DeepSWE | 29.9% | — |
| FrontierCode | 24.3% | — |
Agentic & Tool Use Not comparable
Claude Sonnet 4.6: 39.1 (#26), Gemma 4 31B IT: —
| Benchmark | Claude Sonnet 4.6 | Gemma 4 31B IT |
|---|---|---|
| Terminal-Bench | 53.4% | — |
| APEX-Agents | 43% | — |
| OSWorld 2.0 | 9.3% | — |
| DeepResearch Bench | 54.9% | — |
| OSWorld | 72.1% | — |
| ExploitBench | 23.6% | — |
| GBAEval | 48.8% | — |
| GDP.pdf | 18% | — |
| LMArena Search | 1221 | — |
| Vending-Bench 2 | 7,204 | — |
Reasoning Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 46.1 (#40), Gemma 4 31B IT: 26.4 (#119)
| Benchmark | Claude Sonnet 4.6 | Gemma 4 31B IT |
|---|---|---|
| NYT Connections (extended) | 80.9% | 70.6% |
| CritPt | 3.1% | 1.4% |
| Chess Puzzles | 13% | 5% |
| Thematic Generalization | 76.3% | 53% |
| LMArena Hard Prompts | 1485 | 1448 |
| DTBench | 89.9% | 82.7% |
| LMCA | 46.5% | 39.3% |
| Epoch Capabilities Index | 152.25 | 142.74 |
| ARC-AGI-2 | 60.4% | — |
| Kagi LLM Benchmark | — | 63.5% |
| ARC-AGI-1 | 86.5% | — |
| Mystery Game Puzzles | 16% | — |
| Surface Evolver Bench | — | 30.6% |
| ForecastBench | 62 | — |
Math Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 51.9 (#41), Gemma 4 31B IT: 41.3 (#72)
| Benchmark | Claude Sonnet 4.6 | Gemma 4 31B IT |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 85.8% | 73.3% |
| LMArena Math | 1463 | 1464 |
| ProofBench | 45% | — |
| FrontierMath (Feb 2025 set) | 32.4% | — |
| FrontierMath Tier 4 (v1) | 8.3% | — |
Knowledge Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 51.6 (#60), Gemma 4 31B IT: 37.9 (#141)
| Benchmark | Claude Sonnet 4.6 | Gemma 4 31B IT |
|---|---|---|
| GPQA Diamond | 87.4% | 75.8% |
| SimpleQA Verified | 35.5% | 10.4% |
| Vectara Hallucination Rate | 10.6% | 7.4% |
| LMArena Expert | 1500 | 1465 |
Multimodal Gemma 4 31B IT leads
Claude Sonnet 4.6: 38.0 (#61), Gemma 4 31B IT: 41.0 (#32)
| Benchmark | Claude Sonnet 4.6 | Gemma 4 31B IT |
|---|---|---|
| LMArena Vision | 1283 | 1278 |
| LMArena Document | 1482 | 1425 |
| Blueprint-Bench 2 | 6.7% | — |
Multilingual Too close to call
Claude Sonnet 4.6: 54.6 (#39), Gemma 4 31B IT: 54.0 (#52)
| Benchmark | Claude Sonnet 4.6 | Gemma 4 31B IT |
|---|---|---|
| LMArena Non-English | 1440 | 1431 |
| LMArena Chinese | 1491 | 1478 |
| LMArena French | 1465 | 1436 |
| LMArena Russian | 1440 | 1462 |
| LMArena Spanish | 1465 | 1443 |
| LMArena German | 1428 | — |
| LMArena Japanese | 1419 | — |
| LMArena Korean | 1411 | — |
Instruction Following Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 77.6 (#23), Gemma 4 31B IT: 75.6 (#58)
| Benchmark | Claude Sonnet 4.6 | Gemma 4 31B IT |
|---|---|---|
| LMArena Instruction Following | 1476 | 1433 |
Long Context Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 45.3 (#40), Gemma 4 31B IT: 44.2 (#67)
| Benchmark | Claude Sonnet 4.6 | Gemma 4 31B IT |
|---|---|---|
| LMArena Longer Query | 1479 | 1446 |
Writing & Preference Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 70.4 (#22), Gemma 4 31B IT: 60.7 (#92)
| Benchmark | Claude Sonnet 4.6 | Gemma 4 31B IT |
|---|---|---|
| LMArena Text | 1458 | 1443 |
| LMArena Creative Writing | 1435 | 1416 |
| EQ-Bench Creative Writing | 1810 | 1368 |
| EQ-Bench 4 | 1207 | 1120 |
| LMArena Multi-Turn | 1464 | 1452 |
Frequently asked questions
Is Claude Sonnet 4.6 better than Gemma 4 31B IT?
Claude Sonnet 4.6 is the stronger model overall, scoring 50.1 to 43.2 on the Noometry Index. Gemma 4 31B IT costs 39× less per token, which makes it the better buy when Claude Sonnet 4.6's lead doesn't matter for your workload.
Which is cheaper, Claude Sonnet 4.6 or Gemma 4 31B IT?
Gemma 4 31B IT is cheaper. It lists at $0.09 per million input tokens and $0.34 per million output tokens; Claude Sonnet 4.6 lists at $3 and $15.
Is Claude Sonnet 4.6 or Gemma 4 31B IT better for coding?
Claude Sonnet 4.6 scores higher on coding benchmarks: 46.0 versus 41.9 in the Noometry coding category.
Which has the bigger context window?
Claude Sonnet 4.6 does, with 1M tokens against 262K.
How many benchmarks do Claude Sonnet 4.6 and Gemma 4 31B IT share?
33 benchmarks have published results for both models. Claude Sonnet 4.6 has 57 scored results on Noometry and Gemma 4 31B IT has 35.