# DeepSeek Coder 33B vs GPT-4 Turbo

> GPT-4 Turbo has enough public results to be ranked (#292); DeepSeek Coder 33B does not yet, so treat this comparison as directional.

- Canonical page: https://noometry.com/compare/deepseek-coder-33b-vs-gpt-4-turbo
- Last updated: 2026-10-11
- Shared benchmarks: 6

## Summary

- They share 6 benchmarks with published results for both. DeepSeek Coder 33B scores higher in 1 category and GPT-4 Turbo in 0 categories; one gap is clear of the uncertainty.
- The widest gap is in coding, where DeepSeek Coder 33B leads 38.0 to 33.8.
- The biggest single-benchmark swing is BigCodeBench Complete: 51.1% for DeepSeek Coder 33B and 58.2% for GPT-4 Turbo.
- DeepSeek Coder 33B has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek Coder 33B | GPT-4 Turbo |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 38.9 | 30.5 |
| Rank | — | 292 |
| Context | — | 128K |
| Input $/M | — | $10 |
| Output $/M | — | $30 |
| Weights | Open | Proprietary |

## Coding

- DeepSeek Coder 33B: 38.0 (#184)
- GPT-4 Turbo: 33.8 (#249)

| Benchmark | DeepSeek Coder 33B | GPT-4 Turbo |
|---|---|---|
| BigCodeBench Instruct | 42% | 48.2% |
| BigCodeBench Complete | 51.1% | 58.2% |
| HumanEval+ | 75% | 86.6% |
| MBPP+ | 70.1% | 73.3% |
| WeirdML | — | 18% |
| LMArena Coding | — | 1268 |

## Agentic & Tool Use

- DeepSeek Coder 33B: —
- GPT-4 Turbo: —

| Benchmark | DeepSeek Coder 33B | GPT-4 Turbo |
|---|---|---|
| METR Time Horizons | — | 36.7% |

## Reasoning

- DeepSeek Coder 33B: —
- GPT-4 Turbo: 15.3 (#317)

| Benchmark | DeepSeek Coder 33B | GPT-4 Turbo |
|---|---|---|
| Epoch Capabilities Index | 96.32 | 127.25 |
| SimpleBench | — | 25.1% |
| Chess Puzzles | — | 6% |
| LMArena Hard Prompts | — | 1251 |
| DTBench | — | 61.6% |
| LMCA | — | 9.8% |
| ForecastBench | — | 59.4 |
| WinoGrande | 62% | — |

## Math

- DeepSeek Coder 33B: —
- GPT-4 Turbo: 9.0 (#322)

| Benchmark | DeepSeek Coder 33B | GPT-4 Turbo |
|---|---|---|
| FrontierMath (Tiers 1-3) | — | 0.7% |
| OTIS Mock AIME 2024-2025 | — | 6.7% |
| LMArena Math | — | 1272 |
| MATH Level 5 | — | 46.7% |
| GSM8K | 35.4% | — |

## Knowledge

- DeepSeek Coder 33B: —
- GPT-4 Turbo: 24.3 (#268)

| Benchmark | DeepSeek Coder 33B | GPT-4 Turbo |
|---|---|---|
| MMLU | 39.4% | 81.3% |
| GPQA Diamond | — | 46.6% |
| Confabulations | — | 28.4% |
| LMArena Expert | — | 1223 |
| ARC (AI2) Challenge | 42.2% | — |

## Multimodal

- DeepSeek Coder 33B: —
- GPT-4 Turbo: 30.6 (#110)

| Benchmark | DeepSeek Coder 33B | GPT-4 Turbo |
|---|---|---|
| LMArena Vision | — | 1090 |

## Multilingual

- DeepSeek Coder 33B: —
- GPT-4 Turbo: 40.5 (#216)

| Benchmark | DeepSeek Coder 33B | GPT-4 Turbo |
|---|---|---|
| LMArena Non-English | — | 1245 |
| LMArena Chinese | — | 1242 |
| LMArena French | — | 1276 |
| LMArena German | — | 1259 |
| LMArena Japanese | — | 1194 |
| LMArena Korean | — | 1187 |
| LMArena Russian | — | 1259 |
| LMArena Spanish | — | 1260 |

## Instruction Following

- DeepSeek Coder 33B: —
- GPT-4 Turbo: 65.8 (#216)

| Benchmark | DeepSeek Coder 33B | GPT-4 Turbo |
|---|---|---|
| LMArena Instruction Following | — | 1249 |

## Long Context

- DeepSeek Coder 33B: —
- GPT-4 Turbo: 38.0 (#206)

| Benchmark | DeepSeek Coder 33B | GPT-4 Turbo |
|---|---|---|
| LMArena Longer Query | — | 1254 |

## Writing & Preference

- DeepSeek Coder 33B: —
- GPT-4 Turbo: 47.7 (#206)

| Benchmark | DeepSeek Coder 33B | GPT-4 Turbo |
|---|---|---|
| LMArena Text | — | 1272 |
| LMArena Creative Writing | — | 1269 |
| LMArena Multi-Turn | — | 1267 |

## FAQ

### Is DeepSeek Coder 33B better than GPT-4 Turbo?

GPT-4 Turbo has enough public results to be ranked (#292); DeepSeek Coder 33B does not yet, so treat this comparison as directional.

### Is DeepSeek Coder 33B or GPT-4 Turbo better for coding?

DeepSeek Coder 33B scores higher on coding benchmarks: 38.0 versus 33.8 in the Noometry coding category.

### How many benchmarks do DeepSeek Coder 33B and GPT-4 Turbo share?

6 benchmarks have published results for both models. DeepSeek Coder 33B has 9 scored results on Noometry and GPT-4 Turbo has 36.