# Llama 2-70B vs Qwen3 Max

> Qwen3 Max is the stronger model overall, scoring 43.7 to 24.4 on the Noometry Index.

- Canonical page: https://noometry.com/compare/llama-2-70b-vs-qwen3-max
- Last updated: 2026-10-10
- Shared benchmarks: 22

## Summary

- They share 22 benchmarks with published results for both. Llama 2-70B scores higher in 0 categories and Qwen3 Max in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Qwen3 Max leads 48.1 to 7.4.
- The biggest single-benchmark swing is MATH Level 5: 3.3% for Llama 2-70B and 97.1% for Qwen3 Max.
- Llama 2-70B has downloadable open weights; the other is API-only.

## Snapshot

| | Llama 2-70B | Qwen3 Max |
|---|---|---|
| Provider | Meta | Alibaba (Qwen) |
| Noometry Index | 24.4 | 43.7 |
| Rank | 349 | 87 |
| Context | — | 262K |
| Input $/M | — | $1.20 |
| Output $/M | — | $6 |
| Weights | Open | Proprietary |

## Coding

- Llama 2-70B: 31.4 (#286)
- Qwen3 Max: 43.0 (#93)

| Benchmark | Llama 2-70B | Qwen3 Max |
|---|---|---|
| LMArena Coding | 1079 | 1456 |
| ALE-Bench | — | 370.45 |

## Agentic & Tool Use

- Llama 2-70B: —
- Qwen3 Max: —

| Benchmark | Llama 2-70B | Qwen3 Max |
|---|---|---|
| Vending-Bench 2 | — | 71.56 |

## Reasoning

- Llama 2-70B: 14.4 (#325)
- Qwen3 Max: 22.6 (#190)

| Benchmark | Llama 2-70B | Qwen3 Max |
|---|---|---|
| LMArena Hard Prompts | 1073 | 1448 |
| DTBench | 41.6% | 82.1% |
| Epoch Capabilities Index | 113.79 | 142.38 |
| Kagi LLM Benchmark | — | 72.5% |
| NYT Connections (extended) | — | 30.1% |
| Chess Puzzles | — | 4% |
| Mystery Game Puzzles | — | 5% |
| LMCA | — | 28.3% |
| 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

- Llama 2-70B: 8.1 (#326)
- Qwen3 Max: 38.7 (#131)

| Benchmark | Llama 2-70B | Qwen3 Max |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 0% | 73.3% |
| LMArena Math | 1091 | 1446 |
| MATH Level 5 | 3.3% | 97.1% |
| FrontierMath (Tiers 1-3) | — | 18.9% |
| GSM8K | 69.6% | — |

## Knowledge

- Llama 2-70B: 7.4 (#310)
- Qwen3 Max: 48.1 (#78)

| Benchmark | Llama 2-70B | Qwen3 Max |
|---|---|---|
| GPQA Diamond | 26.3% | 72.6% |
| LMArena Expert | 1039 | 1455 |
| SimpleQA Verified | — | 48.7% |
| ARC (AI2) Challenge | 78.3% | — |
| BoolQ | 88.6% | — |
| MMLU | 69.9% | — |
| OpenBookQA | 60.2% | — |
| TriviaQA | 87.6% | — |

## Multilingual

- Llama 2-70B: 27.7 (#274)
- Qwen3 Max: 53.7 (#62)

| Benchmark | Llama 2-70B | Qwen3 Max |
|---|---|---|
| LMArena Non-English | 1045 | 1429 |
| LMArena Chinese | 995 | 1478 |
| LMArena French | 1090 | 1449 |
| LMArena German | 1041 | 1463 |
| LMArena Japanese | 927 | 1397 |
| LMArena Korean | 964 | 1399 |
| LMArena Russian | 1083 | 1428 |
| LMArena Spanish | 1143 | 1462 |

## Instruction Following

- Llama 2-70B: 54.9 (#278)
- Qwen3 Max: 74.8 (#87)

| Benchmark | Llama 2-70B | Qwen3 Max |
|---|---|---|
| LMArena Instruction Following | 1071 | 1419 |

## Long Context

- Llama 2-70B: 32.3 (#270)
- Qwen3 Max: 41.6 (#134)

| Benchmark | Llama 2-70B | Qwen3 Max |
|---|---|---|
| LMArena Longer Query | 1062 | 1438 |
| Fiction.LiveBench | — | 66.7% |
| CL-bench | — | 14.5% |

## Writing & Preference

- Llama 2-70B: 32.3 (#279)
- Qwen3 Max: 62.4 (#76)

| Benchmark | Llama 2-70B | Qwen3 Max |
|---|---|---|
| LMArena Text | 1115 | 1439 |
| LMArena Creative Writing | 1075 | 1402 |
| LMArena Multi-Turn | 1088 | 1446 |

## FAQ

### Is Llama 2-70B better than Qwen3 Max?

Qwen3 Max is the stronger model overall, scoring 43.7 to 24.4 on the Noometry Index.

### Is Llama 2-70B or Qwen3 Max better for coding?

Qwen3 Max scores higher on coding benchmarks: 43.0 versus 31.4 in the Noometry coding category.

### How many benchmarks do Llama 2-70B and Qwen3 Max share?

22 benchmarks have published results for both models. Llama 2-70B has 35 scored results on Noometry and Qwen3 Max has 33.