# DeepSeek-V3.1-Terminus vs GPT-4

> DeepSeek-V3.1-Terminus is the stronger model overall, scoring 43.1 to 29.1 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v3-1-terminus-vs-gpt-4
- Last updated: 2026-10-10
- Shared benchmarks: 12

## Summary

- They share 12 benchmarks with published results for both. DeepSeek-V3.1-Terminus scores higher in 7 categories and GPT-4 in 0 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-V3.1-Terminus leads 38.5 to 10.8.
- The biggest single-benchmark swing is DTBench: 81.3% for DeepSeek-V3.1-Terminus and 62.7% for GPT-4.
- DeepSeek-V3.1-Terminus is cheaper at $0.27 / $1 per million input/output tokens, against $30 / $60 for GPT-4.
- DeepSeek-V3.1-Terminus accepts more context: 164K tokens versus 8K.
- DeepSeek-V3.1-Terminus has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek-V3.1-Terminus | GPT-4 |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 43.1 | 29.1 |
| Rank | 97 | 316 |
| Context | 164K | 8K |
| Input $/M | $0.27 | $30 |
| Output $/M | $1 | $60 |
| Weights | Open | Proprietary |

## Coding

- DeepSeek-V3.1-Terminus: 42.0 (#113)
- GPT-4: 31.6 (#283)

| Benchmark | DeepSeek-V3.1-Terminus | GPT-4 |
|---|---|---|
| LMArena Coding | 1426 | 1254 |
| SciCode | 40.6% | — |
| WeirdML | — | 12.4% |
| BigCodeBench Instruct | — | 46% |
| BigCodeBench Complete | — | 57.2% |
| ALE-Bench | 745.17 | — |
| HumanEval+ | — | 79.3% |

## Agentic & Tool Use

- DeepSeek-V3.1-Terminus: —
- GPT-4: —

| Benchmark | DeepSeek-V3.1-Terminus | GPT-4 |
|---|---|---|
| METR Time Horizons | — | 36.1% |

## Reasoning

- DeepSeek-V3.1-Terminus: 26.4 (#133)
- GPT-4: 17.8 (#289)

| Benchmark | DeepSeek-V3.1-Terminus | GPT-4 |
|---|---|---|
| LMArena Hard Prompts | 1426 | 1241 |
| DTBench | 81.3% | 62.7% |
| LMCA | 28.6% | 17.1% |
| Kagi LLM Benchmark | 57.4% | — |
| CritPt | 1.7% | — |
| Chess Puzzles | — | 4% |
| Mystery Game Puzzles | — | 12% |
| BIG-Bench Hard | — | 75.1% |
| Epoch Capabilities Index | — | 125.89 |
| ForecastBench | — | 57.8 |
| HellaSwag | — | 95.3% |
| WinoGrande | — | 87.5% |

## Math

- DeepSeek-V3.1-Terminus: 38.5 (#137)
- GPT-4: 10.8 (#309)

| Benchmark | DeepSeek-V3.1-Terminus | GPT-4 |
|---|---|---|
| LMArena Math | 1402 | 1269 |
| OTIS Mock AIME 2024-2025 | — | 1.1% |
| MATH Level 5 | — | 23% |
| GSM8K | — | 92% |

## Knowledge

- DeepSeek-V3.1-Terminus: —
- GPT-4: 18.4 (#282)

| Benchmark | DeepSeek-V3.1-Terminus | GPT-4 |
|---|---|---|
| GPQA Diamond | — | 35.7% |
| LMArena Expert | — | 1211 |
| MMLU | — | 86.4% |
| TriviaQA | — | 84.8% |

## Multilingual

- DeepSeek-V3.1-Terminus: 52.1 (#92)
- GPT-4: 40.6 (#215)

| Benchmark | DeepSeek-V3.1-Terminus | GPT-4 |
|---|---|---|
| LMArena Non-English | 1407 | 1246 |
| LMArena Russian | 1436 | 1251 |
| LMArena Chinese | — | 1242 |
| LMArena French | — | 1283 |
| LMArena German | — | 1251 |
| LMArena Japanese | — | 1209 |
| LMArena Korean | — | 1184 |
| LMArena Spanish | — | 1261 |

## Instruction Following

- DeepSeek-V3.1-Terminus: 74.0 (#106)
- GPT-4: 65.3 (#222)

| Benchmark | DeepSeek-V3.1-Terminus | GPT-4 |
|---|---|---|
| LMArena Instruction Following | 1404 | 1241 |

## Long Context

- DeepSeek-V3.1-Terminus: 43.4 (#97)
- GPT-4: 37.7 (#212)

| Benchmark | DeepSeek-V3.1-Terminus | GPT-4 |
|---|---|---|
| LMArena Longer Query | 1421 | 1244 |

## Writing & Preference

- DeepSeek-V3.1-Terminus: 61.0 (#92)
- GPT-4: 34.9 (#268)

| Benchmark | DeepSeek-V3.1-Terminus | GPT-4 |
|---|---|---|
| LMArena Text | 1419 | 1263 |
| LMArena Creative Writing | 1403 | 1244 |
| LMArena Multi-Turn | 1411 | 1257 |
| EQ-Bench Creative Writing | — | 752 |

## FAQ

### Is DeepSeek-V3.1-Terminus better than GPT-4?

DeepSeek-V3.1-Terminus is the stronger model overall, scoring 43.1 to 29.1 on the Noometry Index.

### Which is cheaper, DeepSeek-V3.1-Terminus or GPT-4?

DeepSeek-V3.1-Terminus is cheaper. It lists at $0.27 per million input tokens and $1 per million output tokens; GPT-4 lists at $30 and $60.

### Is DeepSeek-V3.1-Terminus or GPT-4 better for coding?

DeepSeek-V3.1-Terminus scores higher on coding benchmarks: 42.0 versus 31.6 in the Noometry coding category.

### Which has the bigger context window?

DeepSeek-V3.1-Terminus does, with 164K tokens against 8K.

### How many benchmarks do DeepSeek-V3.1-Terminus and GPT-4 share?

12 benchmarks have published results for both models. DeepSeek-V3.1-Terminus has 16 scored results on Noometry and GPT-4 has 38.