# Gemma 4 31B IT vs Llama 2-70B

> Gemma 4 31B IT is the stronger model overall, scoring 43.5 to 24.4 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gemma-4-31b-it-vs-llama-2-70b
- Last updated: 2026-10-12
- Shared benchmarks: 18

## Summary

- They share 18 benchmarks with published results for both. Gemma 4 31B IT scores higher in 8 categories and Llama 2-70B in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Gemma 4 31B IT leads 43.2 to 8.1.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 73.3% for Gemma 4 31B IT and 0% for Llama 2-70B.

## Snapshot

| | Gemma 4 31B IT | Llama 2-70B |
|---|---|---|
| Provider | Google | Meta |
| Noometry Index | 43.5 | 24.4 |
| Rank | 90 | 349 |
| Context | 262K | — |
| Input $/M | $0.09 | — |
| Output $/M | $0.34 | — |
| Weights | Open | Open |

## Coding

- Gemma 4 31B IT: 42.3 (#108)
- Llama 2-70B: 31.4 (#286)

| Benchmark | Gemma 4 31B IT | Llama 2-70B |
|---|---|---|
| LMArena Coding | 1459 | 1079 |
| LMArena WebDev | 1366 | — |
| SciCode | 43.4% | — |
| WeirdML | 52.3% | — |
| ALE-Bench | 925.5 | — |

## Reasoning

- Gemma 4 31B IT: 27.2 (#122)
- Llama 2-70B: 14.4 (#325)

| Benchmark | Gemma 4 31B IT | Llama 2-70B |
|---|---|---|
| LMArena Hard Prompts | 1448 | 1073 |
| DTBench | 82.7% | 41.6% |
| Epoch Capabilities Index | 142.74 | 113.79 |
| Kagi LLM Benchmark | 63.5% | — |
| NYT Connections (extended) | 70.6% | — |
| CritPt | 1.4% | — |
| Chess Puzzles | 5% | — |
| Thematic Generalization | 53% | — |
| LMCA | 39.3% | — |
| Surface Evolver Bench | 30.6% | — |
| BIG-Bench Hard | — | 64.9% |
| CommonsenseQA 2.0 | — | 50% |
| ForecastBench | — | 51.4 |
| HellaSwag | — | 85.3% |
| LAMBADA | — | 78.9% |
| PIQA | — | 82.8% |
| WinoGrande | — | 80.2% |

## Math

- Gemma 4 31B IT: 43.2 (#81)
- Llama 2-70B: 8.1 (#326)

| Benchmark | Gemma 4 31B IT | Llama 2-70B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 73.3% | 0% |
| LMArena Math | 1465 | 1091 |
| MATH Level 5 | — | 3.3% |
| GSM8K | — | 69.6% |

## Knowledge

- Gemma 4 31B IT: 37.9 (#151)
- Llama 2-70B: 7.4 (#310)

| Benchmark | Gemma 4 31B IT | Llama 2-70B |
|---|---|---|
| GPQA Diamond | 75.8% | 26.3% |
| LMArena Expert | 1465 | 1039 |
| SimpleQA Verified | 10.4% | — |
| Vectara Hallucination Rate | 7.4% | — |
| ARC (AI2) Challenge | — | 78.3% |
| BoolQ | — | 88.6% |
| MMLU | — | 69.9% |
| OpenBookQA | — | 60.2% |
| TriviaQA | — | 87.6% |

## Multimodal

- Gemma 4 31B IT: 41.6 (#34)
- Llama 2-70B: —

| Benchmark | Gemma 4 31B IT | Llama 2-70B |
|---|---|---|
| LMArena Vision | 1277 | — |
| LMArena Document | 1425 | — |

## Multilingual

- Gemma 4 31B IT: 53.8 (#57)
- Llama 2-70B: 27.7 (#274)

| Benchmark | Gemma 4 31B IT | Llama 2-70B |
|---|---|---|
| LMArena Non-English | 1431 | 1045 |
| LMArena Chinese | 1476 | 995 |
| LMArena French | 1435 | 1090 |
| LMArena Russian | 1460 | 1083 |
| LMArena Spanish | 1444 | 1143 |
| LMArena German | — | 1041 |
| LMArena Japanese | — | 927 |
| LMArena Korean | — | 964 |

## Instruction Following

- Gemma 4 31B IT: 75.5 (#61)
- Llama 2-70B: 54.9 (#278)

| Benchmark | Gemma 4 31B IT | Llama 2-70B |
|---|---|---|
| LMArena Instruction Following | 1433 | 1071 |

## Long Context

- Gemma 4 31B IT: 44.2 (#71)
- Llama 2-70B: 32.3 (#270)

| Benchmark | Gemma 4 31B IT | Llama 2-70B |
|---|---|---|
| LMArena Longer Query | 1446 | 1062 |

## Writing & Preference

- Gemma 4 31B IT: 60.5 (#96)
- Llama 2-70B: 32.3 (#279)

| Benchmark | Gemma 4 31B IT | Llama 2-70B |
|---|---|---|
| LMArena Text | 1443 | 1115 |
| LMArena Creative Writing | 1415 | 1075 |
| LMArena Multi-Turn | 1452 | 1088 |
| EQ-Bench Creative Writing | 1368 | — |
| EQ-Bench 4 | 1120 | — |

## FAQ

### Is Gemma 4 31B IT better than Llama 2-70B?

Gemma 4 31B IT is the stronger model overall, scoring 43.5 to 24.4 on the Noometry Index.

### Is Gemma 4 31B IT or Llama 2-70B better for coding?

Gemma 4 31B IT scores higher on coding benchmarks: 42.3 versus 31.4 in the Noometry coding category.

### How many benchmarks do Gemma 4 31B IT and Llama 2-70B share?

18 benchmarks have published results for both models. Gemma 4 31B IT has 35 scored results on Noometry and Llama 2-70B has 35.