# Hy4 preview vs Llama 2-70B

> Hy4 preview is the stronger model overall, scoring 44.6 to 24.1 on the Noometry Index.

- Canonical page: https://noometry.com/compare/hy4-preview-vs-llama-2-70b
- Last updated: 2026-10-10
- Shared benchmarks: 0

## Summary

- The widest gap is in math, where Hy4 preview leads 54.5 to 7.5.

## Snapshot

| | Hy4 preview | Llama 2-70B |
|---|---|---|
| Provider | Tencent | Meta |
| Noometry Index | 44.6 | 24.1 |
| Rank | 73 | 343 |
| Context | 1.05M | — |
| Input $/M | $0.83 | — |
| Output $/M | $2.50 | — |
| Weights | Open | Open |

## Coding

- Hy4 preview: 51.4 (#36)
- Llama 2-70B: 30.8 (#281)

| Benchmark | Hy4 preview | Llama 2-70B |
|---|---|---|
| LMArena WebDev | 1633 | — |
| LMArena Coding | — | 1079 |

## Reasoning

- Hy4 preview: 30.6 (#75)
- Llama 2-70B: 13.7 (#324)

| Benchmark | Hy4 preview | Llama 2-70B |
|---|---|---|
| NYT Connections (extended) | 68.2% | — |
| LMArena Hard Prompts | — | 1073 |
| DTBench | — | 41.6% |
| BIG-Bench Hard | — | 64.9% |
| CommonsenseQA 2.0 | — | 50% |
| Epoch Capabilities Index | — | 113.59 |
| ForecastBench | — | 51.4 |
| HellaSwag | — | 85.3% |
| LAMBADA | — | 78.9% |
| PIQA | — | 82.8% |
| WinoGrande | — | 80.2% |

## Math

- Hy4 preview: 54.5 (#37)
- Llama 2-70B: 7.5 (#320)

| Benchmark | Hy4 preview | Llama 2-70B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 0% |
| ProofBench | 75% | — |
| LMArena Math | — | 1091 |
| MATH Level 5 | — | 3.3% |
| GSM8K | — | 69.6% |

## Knowledge

- Hy4 preview: —
- Llama 2-70B: 7.4 (#305)

| Benchmark | Hy4 preview | Llama 2-70B |
|---|---|---|
| GPQA Diamond | — | 26.3% |
| LMArena Expert | — | 1039 |
| ARC (AI2) Challenge | — | 78.3% |
| BoolQ | — | 88.6% |
| MMLU | — | 69.9% |
| OpenBookQA | — | 60.2% |
| TriviaQA | — | 87.6% |

## Multilingual

- Hy4 preview: —
- Llama 2-70B: 27.9 (#269)

| Benchmark | Hy4 preview | Llama 2-70B |
|---|---|---|
| LMArena Non-English | — | 1045 |
| LMArena Chinese | — | 994 |
| LMArena French | — | 1089 |
| LMArena German | — | 1041 |
| LMArena Japanese | — | 927 |
| LMArena Korean | — | 964 |
| LMArena Russian | — | 1083 |
| LMArena Spanish | — | 1142 |

## Instruction Following

- Hy4 preview: —
- Llama 2-70B: 55.1 (#273)

| Benchmark | Hy4 preview | Llama 2-70B |
|---|---|---|
| LMArena Instruction Following | — | 1071 |

## Long Context

- Hy4 preview: —
- Llama 2-70B: 32.2 (#265)

| Benchmark | Hy4 preview | Llama 2-70B |
|---|---|---|
| LMArena Longer Query | — | 1062 |

## Writing & Preference

- Hy4 preview: —
- Llama 2-70B: 32.5 (#274)

| Benchmark | Hy4 preview | Llama 2-70B |
|---|---|---|
| LMArena Text | — | 1115 |
| LMArena Creative Writing | — | 1075 |
| LMArena Multi-Turn | — | 1088 |

## FAQ

### Is Hy4 preview better than Llama 2-70B?

Hy4 preview is the stronger model overall, scoring 44.6 to 24.1 on the Noometry Index.

### Is Hy4 preview or Llama 2-70B better for coding?

Hy4 preview scores higher on coding benchmarks: 51.4 versus 30.8 in the Noometry coding category.

### How many benchmarks do Hy4 preview and Llama 2-70B share?

0 benchmarks have published results for both models. Hy4 preview has 3 scored results on Noometry and Llama 2-70B has 35.