# Claude Sonnet 4.5 vs Gemma 3 4B

> Claude Sonnet 4.5 is the stronger model overall, scoring 44.0 to 27.9 on the Noometry Index. Gemma 3 4B costs 120× less per token, which makes it the better buy when Claude Sonnet 4.5's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/claude-sonnet-4-5-vs-gemma-3-4b
- Last updated: 2026-10-10
- Shared benchmarks: 22

## Summary

- They share 22 benchmarks with published results for both. Claude Sonnet 4.5 scores higher in 9 categories and Gemma 3 4B in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Claude Sonnet 4.5 leads 48.4 to 11.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 77.8% for Claude Sonnet 4.5 and 7.5% for Gemma 3 4B.
- Gemma 3 4B is cheaper at $0.04 / $0.08 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.5.
- Claude Sonnet 4.5 accepts more context: 1M tokens versus 131K.
- Gemma 3 4B has downloadable open weights; the other is API-only.

## Snapshot

| | Claude Sonnet 4.5 | Gemma 3 4B |
|---|---|---|
| Provider | Anthropic | Google |
| Noometry Index | 44.0 | 27.9 |
| Rank | 76 | 320 |
| Context | 1M | 131K |
| Input $/M | $3 | $0.04 |
| Output $/M | $15 | $0.08 |
| Weights | Proprietary | Open |

## Coding

- Claude Sonnet 4.5: 47.8 (#54)
- Gemma 3 4B: 35.3 (#214)

| Benchmark | Claude Sonnet 4.5 | Gemma 3 4B |
|---|---|---|
| LMArena Coding | 1489 | 1230 |
| SWE-bench Verified | 71.3% | — |
| SWE-bench Verified (bash only) | 71.4% | — |
| LMArena WebDev | 1393 | — |
| SWE-bench Multilingual | 67% | — |
| SciCode | 44.7% | — |
| GSO | 14.7% | — |
| WeirdML | 47.7% | — |
| ALE-Bench | 796.15 | — |
| AlgoTune | 1.52 | — |

## Agentic & Tool Use

- Claude Sonnet 4.5: 38.1 (#30)
- Gemma 3 4B: 20.7 (#134)

| Benchmark | Claude Sonnet 4.5 | Gemma 3 4B |
|---|---|---|
| Berkeley Function Calling Leaderboard | 73.2% | 19.6% |
| Terminal-Bench | 46.5% | — |
| GDPval | 42.5% | — |
| Remote Labor Index | 2.1% | — |
| τ²-bench Airline | 72% | — |
| τ²-bench Banking | 25.3% | — |
| τ²-bench Retail | 72.4% | — |
| τ²-bench Telecom | 84.9% | — |
| Cybench | 60% | — |
| DeepResearch Bench | 52.6% | — |
| OSWorld | 62.9% | — |
| LMArena Search | 1159 | — |
| METR Time Horizons | 67.4% | — |
| Vending-Bench 2 | 3,839 | — |

## Reasoning

- Claude Sonnet 4.5: 26.9 (#113)
- Gemma 3 4B: 12.7 (#330)

| Benchmark | Claude Sonnet 4.5 | Gemma 3 4B |
|---|---|---|
| Kagi LLM Benchmark | 57.9% | 25.2% |
| Chess Puzzles | 12% | 0% |
| LMArena Hard Prompts | 1462 | 1253 |
| DTBench | 83.2% | 50.9% |
| LMCA | 38.8% | 2.8% |
| Epoch Capabilities Index | 146.84 | 116.02 |
| ARC-AGI-2 | 13.6% | — |
| SimpleBench | 54.3% | — |
| NYT Connections (extended) | 37.3% | — |
| ARC-AGI-1 | 63.7% | — |
| CritPt | 1.1% | — |
| EnigmaEval | 6% | — |
| EBR-Bench | 2.4% | — |
| Mystery Game Puzzles | 17% | — |
| ForecastBench | 61.9 | — |

## Math

- Claude Sonnet 4.5: 31.9 (#190)
- Gemma 3 4B: 15.8 (#286)

| Benchmark | Claude Sonnet 4.5 | Gemma 3 4B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 77.8% | 7.5% |
| LMArena Math | 1448 | 1239 |
| FrontierMath (Tiers 1-3) | 23.9% | — |
| FrontierMath Tier 4 | 2.4% | — |
| ProofBench | 19% | — |
| Omni-MATH | 55.3% | — |
| MATH Level 5 | 97.7% | — |
| FrontierMath (Feb 2025 set) | 15.2% | — |
| FrontierMath Tier 4 (v1) | 4.2% | — |

## Knowledge

- Claude Sonnet 4.5: 48.4 (#70)
- Gemma 3 4B: 11.8 (#294)

| Benchmark | Claude Sonnet 4.5 | Gemma 3 4B |
|---|---|---|
| GPQA Diamond | 82.3% | 23.2% |
| Vectara Hallucination Rate | 12% | 6.4% |
| LMArena Expert | 1483 | 1223 |
| Humanity's Last Exam | 13.7% | — |
| SimpleQA Verified | 30.7% | — |
| MMLU-Pro | 86.9% | — |
| GPQA (HELM) | 68.6% | — |

## Multimodal

- Claude Sonnet 4.5: 34.1 (#84)
- Gemma 3 4B: —

| Benchmark | Claude Sonnet 4.5 | Gemma 3 4B |
|---|---|---|
| VPCT | 39.8% | — |
| LMArena Document | 1450 | — |

## Multilingual

- Claude Sonnet 4.5: 53.6 (#65)
- Gemma 3 4B: 42.7 (#189)

| Benchmark | Claude Sonnet 4.5 | Gemma 3 4B |
|---|---|---|
| LMArena Non-English | 1425 | 1273 |
| LMArena German | 1427 | 1281 |
| LMArena Russian | 1436 | 1294 |
| LMArena Chinese | 1459 | — |
| LMArena French | 1459 | — |
| LMArena Japanese | 1390 | — |
| LMArena Korean | 1403 | — |
| LMArena Spanish | 1457 | — |

## Instruction Following

- Claude Sonnet 4.5: 75.2 (#74)
- Gemma 3 4B: 65.4 (#219)

| Benchmark | Claude Sonnet 4.5 | Gemma 3 4B |
|---|---|---|
| LMArena Instruction Following | 1459 | 1239 |
| IFEval | 85% | — |

## Long Context

- Claude Sonnet 4.5: 45.2 (#45)
- Gemma 3 4B: 38.6 (#189)

| Benchmark | Claude Sonnet 4.5 | Gemma 3 4B |
|---|---|---|
| LMArena Longer Query | 1476 | 1274 |

## Writing & Preference

- Claude Sonnet 4.5: 66.7 (#33)
- Gemma 3 4B: 42.2 (#234)

| Benchmark | Claude Sonnet 4.5 | Gemma 3 4B |
|---|---|---|
| LMArena Text | 1439 | 1291 |
| LMArena Creative Writing | 1442 | 1270 |
| EQ-Bench Creative Writing | 1678 | 1068 |
| LMArena Multi-Turn | 1465 | 1255 |
| WildBench | 85.4% | — |

## FAQ

### Is Claude Sonnet 4.5 better than Gemma 3 4B?

Claude Sonnet 4.5 is the stronger model overall, scoring 44.0 to 27.9 on the Noometry Index. Gemma 3 4B costs 120× less per token, which makes it the better buy when Claude Sonnet 4.5's lead doesn't matter for your workload.

### Which is cheaper, Claude Sonnet 4.5 or Gemma 3 4B?

Gemma 3 4B is cheaper. It lists at $0.04 per million input tokens and $0.08 per million output tokens; Claude Sonnet 4.5 lists at $3 and $15.

### Is Claude Sonnet 4.5 or Gemma 3 4B better for coding?

Claude Sonnet 4.5 scores higher on coding benchmarks: 47.8 versus 35.3 in the Noometry coding category.

### Which has the bigger context window?

Claude Sonnet 4.5 does, with 1M tokens against 131K.

### How many benchmarks do Claude Sonnet 4.5 and Gemma 3 4B share?

22 benchmarks have published results for both models. Claude Sonnet 4.5 has 73 scored results on Noometry and Gemma 3 4B has 22.