# Deepseek Coder v2 vs DeepSeek-V3.1

> DeepSeek-V3.1 is the stronger model overall, scoring 42.2 to 35.3 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-coder-v2-vs-deepseek-v3-1
- Last updated: 2026-10-10
- Shared benchmarks: 17

## Summary

- They share 17 benchmarks with published results for both. Deepseek Coder v2 scores higher in 1 category and DeepSeek-V3.1 in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V3.1 leads 60.5 to 38.4.

## Snapshot

| | Deepseek Coder v2 | DeepSeek-V3.1 |
|---|---|---|
| Provider | DeepSeek | DeepSeek |
| Noometry Index | 35.3 | 42.2 |
| Rank | 217 | 106 |
| Context | — | 164K |
| Input $/M | — | $0.25 |
| Output $/M | — | $0.95 |
| Weights | Open | Open |

## Coding

- Deepseek Coder v2: 37.5 (#178)
- DeepSeek-V3.1: 39.8 (#140)

| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1 |
|---|---|---|
| LMArena Coding | 1251 | 1417 |
| WeirdML | — | 38.4% |
| BigCodeBench Instruct | 48.2% | — |
| BigCodeBench Complete | 59.7% | — |
| HumanEval+ | 82.3% | — |
| MBPP+ | 75.1% | — |

## Reasoning

- Deepseek Coder v2: 22.6 (#175)
- DeepSeek-V3.1: 27.3 (#109)

| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1 |
|---|---|---|
| LMArena Hard Prompts | 1207 | 1417 |
| SimpleBench | — | 40% |
| Kagi LLM Benchmark | — | 53.2% |
| DTBench | — | 82.7% |
| LMCA | — | 24.3% |
| Epoch Capabilities Index | — | 139.92 |
| ForecastBench | — | 58 |
| WinoGrande | 83.7% | — |

## Math

- Deepseek Coder v2: 32.3 (#188)
- DeepSeek-V3.1: 36.1 (#125)

| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1 |
|---|---|---|
| LMArena Math | 1241 | 1420 |
| GSM8K | 94.5% | — |

## Knowledge

- Deepseek Coder v2: 31.7 (#209)
- DeepSeek-V3.1: 43.0 (#86)

| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1 |
|---|---|---|
| LMArena Expert | 1181 | 1403 |
| Vectara Hallucination Rate | — | 5.5% |
| ARC (AI2) Challenge | 64.3% | — |

## Multilingual

- Deepseek Coder v2: 36.4 (#235)
- DeepSeek-V3.1: 51.8 (#100)

| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1 |
|---|---|---|
| LMArena Non-English | 1182 | 1400 |
| LMArena Chinese | 1201 | 1470 |
| LMArena French | 1184 | 1447 |
| LMArena German | 1164 | 1412 |
| LMArena Japanese | 1126 | 1374 |
| LMArena Korean | 1104 | 1337 |
| LMArena Russian | 1188 | 1405 |
| LMArena Spanish | 1153 | 1429 |

## Instruction Following

- Deepseek Coder v2: 61.9 (#237)
- DeepSeek-V3.1: 74.1 (#105)

| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1 |
|---|---|---|
| LMArena Instruction Following | 1180 | 1400 |

## Long Context

- Deepseek Coder v2: 37.0 (#219)
- DeepSeek-V3.1: 36.3 (#227)

| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1 |
|---|---|---|
| LMArena Longer Query | 1219 | 1422 |
| Fiction.LiveBench | — | 52.8% |

## Writing & Preference

- Deepseek Coder v2: 38.4 (#249)
- DeepSeek-V3.1: 60.5 (#93)

| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1 |
|---|---|---|
| LMArena Text | 1191 | 1420 |
| LMArena Creative Writing | 1120 | 1401 |
| LMArena Multi-Turn | 1176 | 1408 |
| EQ-Bench Creative Writing | — | 1436 |

## FAQ

### Is Deepseek Coder v2 better than DeepSeek-V3.1?

DeepSeek-V3.1 is the stronger model overall, scoring 42.2 to 35.3 on the Noometry Index.

### Is Deepseek Coder v2 or DeepSeek-V3.1 better for coding?

DeepSeek-V3.1 scores higher on coding benchmarks: 39.8 versus 37.5 in the Noometry coding category.

### How many benchmarks do Deepseek Coder v2 and DeepSeek-V3.1 share?

17 benchmarks have published results for both models. Deepseek Coder v2 has 24 scored results on Noometry and DeepSeek-V3.1 has 27.