# Deepseek Coder v2 vs Llama-3.3-70B-Instruct

> Deepseek Coder v2 is the stronger model overall, scoring 35.9 to 30.6 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-coder-v2-vs-llama-3-3-70b-instruct
- Last updated: 2026-10-12
- Shared benchmarks: 19

## Summary

- They share 19 benchmarks with published results for both. Deepseek Coder v2 scores higher in 5 categories and Llama-3.3-70B-Instruct in 3 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Deepseek Coder v2 leads 34.9 to 15.3.

## Snapshot

| | Deepseek Coder v2 | Llama-3.3-70B-Instruct |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 35.9 | 30.6 |
| Rank | 220 | 291 |
| Context | — | 128K |
| Input $/M | — | $0.10 |
| Output $/M | — | $0.32 |
| Weights | Open | Open |

## Coding

- Deepseek Coder v2: 38.1 (#183)
- Llama-3.3-70B-Instruct: 31.0 (#290)

| Benchmark | Deepseek Coder v2 | Llama-3.3-70B-Instruct |
|---|---|---|
| BigCodeBench Instruct | 48.2% | 46.9% |
| LMArena Coding | 1251 | 1268 |
| BigCodeBench Complete | 59.7% | 57.5% |
| SciCode | — | 26% |
| WeirdML | — | 14.4% |
| LiveBench Coding | — | 36.6% |
| HumanEval+ | 82.3% | — |
| MBPP+ | 75.1% | — |

## Agentic & Tool Use

- Deepseek Coder v2: —
- Llama-3.3-70B-Instruct: 25.8 (#105)

| Benchmark | Deepseek Coder v2 | Llama-3.3-70B-Instruct |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 31.9% |
| BALROG | — | 23% |

## Reasoning

- Deepseek Coder v2: 23.6 (#176)
- Llama-3.3-70B-Instruct: 14.1 (#327)

| Benchmark | Deepseek Coder v2 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Hard Prompts | 1207 | 1257 |
| SimpleBench | — | 19.9% |
| CritPt | — | 0% |
| LiveBench Reasoning | — | 50.8% |
| DTBench | — | 59.5% |
| LiveBench Data Analysis | — | 49.5% |
| LMCA | — | 17.5% |
| Epoch Capabilities Index | — | 127.33 |
| ForecastBench | — | 58.6 |
| LiveBench | — | 50.2% |
| WinoGrande | 83.7% | — |

## Math

- Deepseek Coder v2: 34.9 (#190)
- Llama-3.3-70B-Instruct: 15.3 (#298)

| Benchmark | Deepseek Coder v2 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Math | 1241 | 1267 |
| OTIS Mock AIME 2024-2025 | — | 5.1% |
| LiveBench Math | — | 42.2% |
| MATH Level 5 | — | 41.6% |
| GSM8K | 94.5% | — |

## Knowledge

- Deepseek Coder v2: 32.3 (#212)
- Llama-3.3-70B-Instruct: 30.6 (#226)

| Benchmark | Deepseek Coder v2 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Expert | 1181 | 1225 |
| GPQA Diamond | — | 47.4% |
| Confabulations | — | 22.8% |
| Vectara Hallucination Rate | — | 4.1% |
| ARC (AI2) Challenge | 64.3% | — |
| MMLU | — | 86.3% |

## Multilingual

- Deepseek Coder v2: 36.3 (#240)
- Llama-3.3-70B-Instruct: 39.9 (#220)

| Benchmark | Deepseek Coder v2 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Non-English | 1182 | 1236 |
| LMArena Chinese | 1201 | 1217 |
| LMArena French | 1185 | 1281 |
| LMArena German | 1164 | 1251 |
| LMArena Japanese | 1126 | 1150 |
| LMArena Korean | 1104 | 1143 |
| LMArena Russian | 1188 | 1252 |
| LMArena Spanish | 1153 | 1270 |

## Instruction Following

- Deepseek Coder v2: 61.7 (#242)
- Llama-3.3-70B-Instruct: 71.1 (#157)

| Benchmark | Deepseek Coder v2 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Instruction Following | 1180 | 1242 |
| LiveBench Instruction Following | — | 82.7% |

## Long Context

- Deepseek Coder v2: 37.0 (#224)
- Llama-3.3-70B-Instruct: 26.4 (#295)

| Benchmark | Deepseek Coder v2 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Longer Query | 1219 | 1256 |
| Fiction.LiveBench | — | 33.3% |

## Writing & Preference

- Deepseek Coder v2: 38.2 (#253)
- Llama-3.3-70B-Instruct: 47.6 (#207)

| Benchmark | Deepseek Coder v2 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Text | 1191 | 1274 |
| LMArena Creative Writing | 1120 | 1250 |
| LMArena Multi-Turn | 1177 | 1280 |
| LiveBench Language | — | 39.2% |

## FAQ

### Is Deepseek Coder v2 better than Llama-3.3-70B-Instruct?

Deepseek Coder v2 is the stronger model overall, scoring 35.9 to 30.6 on the Noometry Index.

### Is Deepseek Coder v2 or Llama-3.3-70B-Instruct better for coding?

Deepseek Coder v2 scores higher on coding benchmarks: 38.1 versus 31.0 in the Noometry coding category.

### How many benchmarks do Deepseek Coder v2 and Llama-3.3-70B-Instruct share?

19 benchmarks have published results for both models. Deepseek Coder v2 has 24 scored results on Noometry and Llama-3.3-70B-Instruct has 43.