# Devstral Small 2505 vs GPT-4

> Devstral Small 2505 is the stronger model overall, scoring 34.3 to 29.1 on the Noometry Index.

- Canonical page: https://noometry.com/compare/devstral-small-vs-gpt-4
- Last updated: 2026-10-11
- Shared benchmarks: 0

## Summary

- The widest gap is in coding, where Devstral Small 2505 leads 38.9 to 31.6.
- Devstral Small 2505 is cheaper at $0.10 / $0.30 per million input/output tokens, against $30 / $60 for GPT-4.
- Devstral Small 2505 accepts more context: 128K tokens versus 8K.
- Devstral Small 2505 has downloadable open weights; the other is API-only.

## Snapshot

| | Devstral Small 2505 | GPT-4 |
|---|---|---|
| Provider | Mistral AI | OpenAI |
| Noometry Index | 34.3 | 29.1 |
| Rank | 233 | 316 |
| Context | 128K | 8K |
| Input $/M | $0.10 | $30 |
| Output $/M | $0.30 | $60 |
| Weights | Open | Proprietary |

## Coding

- Devstral Small 2505: 38.9 (#166)
- GPT-4: 31.6 (#283)

| Benchmark | Devstral Small 2505 | GPT-4 |
|---|---|---|
| SWE-bench Verified (bash only) | 56.4% | — |
| SciCode | 28.8% | — |
| WeirdML | — | 12.4% |
| BigCodeBench Instruct | — | 46% |
| LMArena Coding | — | 1254 |
| BigCodeBench Complete | — | 57.2% |
| HumanEval+ | — | 79.3% |

## Agentic & Tool Use

- Devstral Small 2505: —
- GPT-4: —

| Benchmark | Devstral Small 2505 | GPT-4 |
|---|---|---|
| METR Time Horizons | — | 36.1% |

## Reasoning

- Devstral Small 2505: 19.7 (#252)
- GPT-4: 17.8 (#289)

| Benchmark | Devstral Small 2505 | GPT-4 |
|---|---|---|
| Kagi LLM Benchmark | 37.7% | — |
| CritPt | 0% | — |
| Chess Puzzles | — | 4% |
| LMArena Hard Prompts | — | 1241 |
| Mystery Game Puzzles | — | 12% |
| DTBench | — | 62.7% |
| LMCA | — | 17.1% |
| BIG-Bench Hard | — | 75.1% |
| Epoch Capabilities Index | — | 125.89 |
| ForecastBench | — | 57.8 |
| HellaSwag | — | 95.3% |
| WinoGrande | — | 87.5% |

## Math

- Devstral Small 2505: —
- GPT-4: 10.8 (#309)

| Benchmark | Devstral Small 2505 | GPT-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 1.1% |
| LMArena Math | — | 1269 |
| MATH Level 5 | — | 23% |
| GSM8K | — | 92% |

## Knowledge

- Devstral Small 2505: —
- GPT-4: 18.4 (#282)

| Benchmark | Devstral Small 2505 | GPT-4 |
|---|---|---|
| GPQA Diamond | — | 35.7% |
| LMArena Expert | — | 1211 |
| MMLU | — | 86.4% |
| TriviaQA | — | 84.8% |

## Multilingual

- Devstral Small 2505: —
- GPT-4: 40.6 (#215)

| Benchmark | Devstral Small 2505 | GPT-4 |
|---|---|---|
| LMArena Non-English | — | 1246 |
| LMArena Chinese | — | 1242 |
| LMArena French | — | 1283 |
| LMArena German | — | 1251 |
| LMArena Japanese | — | 1209 |
| LMArena Korean | — | 1184 |
| LMArena Russian | — | 1251 |
| LMArena Spanish | — | 1261 |

## Instruction Following

- Devstral Small 2505: —
- GPT-4: 65.3 (#222)

| Benchmark | Devstral Small 2505 | GPT-4 |
|---|---|---|
| LMArena Instruction Following | — | 1241 |

## Long Context

- Devstral Small 2505: —
- GPT-4: 37.7 (#212)

| Benchmark | Devstral Small 2505 | GPT-4 |
|---|---|---|
| LMArena Longer Query | — | 1244 |

## Writing & Preference

- Devstral Small 2505: —
- GPT-4: 34.9 (#268)

| Benchmark | Devstral Small 2505 | GPT-4 |
|---|---|---|
| LMArena Text | — | 1263 |
| LMArena Creative Writing | — | 1244 |
| EQ-Bench Creative Writing | — | 752 |
| LMArena Multi-Turn | — | 1257 |

## FAQ

### Is Devstral Small 2505 better than GPT-4?

Devstral Small 2505 is the stronger model overall, scoring 34.3 to 29.1 on the Noometry Index.

### Which is cheaper, Devstral Small 2505 or GPT-4?

Devstral Small 2505 is cheaper. It lists at $0.10 per million input tokens and $0.30 per million output tokens; GPT-4 lists at $30 and $60.

### Is Devstral Small 2505 or GPT-4 better for coding?

Devstral Small 2505 scores higher on coding benchmarks: 38.9 versus 31.6 in the Noometry coding category.

### Which has the bigger context window?

Devstral Small 2505 does, with 128K tokens against 8K.

### How many benchmarks do Devstral Small 2505 and GPT-4 share?

0 benchmarks have published results for both models. Devstral Small 2505 has 4 scored results on Noometry and GPT-4 has 38.