# Devstral Small 2505 vs Qwen3.8 Max

> Qwen3.8 Max is the stronger model overall, scoring 56.8 to 34.3 on the Noometry Index. Devstral Small 2505 costs 20× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/devstral-small-vs-qwen3-8-max
- Last updated: 2026-10-10
- Shared benchmarks: 2

## Summary

- They share 2 benchmarks with published results for both. Devstral Small 2505 scores higher in 0 categories and Qwen3.8 Max in 2 categories; 2 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Qwen3.8 Max leads 54.4 to 19.7.
- The biggest single-benchmark swing is SciCode: 28.8% for Devstral Small 2505 and 53.2% for Qwen3.8 Max.
- Devstral Small 2505 is cheaper at $0.10 / $0.30 per million input/output tokens, against $2 / $6 for Qwen3.8 Max.
- Qwen3.8 Max accepts more context: 1M tokens versus 128K.
- Devstral Small 2505 has downloadable open weights; the other is API-only.

## Snapshot

| | Devstral Small 2505 | Qwen3.8 Max |
|---|---|---|
| Provider | Mistral AI | Alibaba (Qwen) |
| Noometry Index | 34.3 | 56.8 |
| Rank | 233 | 22 |
| Context | 128K | 1M |
| Input $/M | $0.10 | $2 |
| Output $/M | $0.30 | $6 |
| Weights | Open | Proprietary |

## Coding

- Devstral Small 2505: 38.9 (#166)
- Qwen3.8 Max: 53.5 (#29)

| Benchmark | Devstral Small 2505 | Qwen3.8 Max |
|---|---|---|
| SciCode | 28.8% | 53.2% |
| DeepSWE | — | 57.5% |
| SWE-bench Verified (bash only) | 56.4% | — |
| LMArena WebDev | — | 1674 |
| FrontierSWE | — | 17.8% |
| LMArena Coding | — | 1502 |

## Agentic & Tool Use

- Devstral Small 2505: —
- Qwen3.8 Max: 45.4 (#14)

| Benchmark | Devstral Small 2505 | Qwen3.8 Max |
|---|---|---|
| APEX-Agents | — | 63.3% |
| τ²-bench Banking | — | 55.1% |
| GDP.pdf | — | 23.2% |

## Reasoning

- Devstral Small 2505: 19.7 (#252)
- Qwen3.8 Max: 54.4 (#26)

| Benchmark | Devstral Small 2505 | Qwen3.8 Max |
|---|---|---|
| CritPt | 0% | 20% |
| Kagi LLM Benchmark | 37.7% | — |
| NYT Connections (extended) | — | 88.3% |
| Chess Puzzles | — | 40% |
| LMArena Hard Prompts | — | 1496 |
| Mystery Game Puzzles | — | 38% |
| DTBench | — | 92% |
| LMCA | — | 46.2% |
| Epoch Capabilities Index | — | 156.41 |

## Math

- Devstral Small 2505: —
- Qwen3.8 Max: 73.2 (#20)

| Benchmark | Devstral Small 2505 | Qwen3.8 Max |
|---|---|---|
| FrontierMath (Tiers 1-3) | — | 74.7% |
| FrontierMath Tier 4 | — | 46.3% |
| OTIS Mock AIME 2024-2025 | — | 100% |
| ProofBench | — | 58% |
| LMArena Math | — | 1499 |

## Knowledge

- Devstral Small 2505: —
- Qwen3.8 Max: 61.7 (#27)

| Benchmark | Devstral Small 2505 | Qwen3.8 Max |
|---|---|---|
| GPQA Diamond | — | 92.7% |
| SimpleQA Verified | — | 47.3% |
| LMArena Expert | — | 1507 |

## Multimodal

- Devstral Small 2505: —
- Qwen3.8 Max: 37.2 (#75)

| Benchmark | Devstral Small 2505 | Qwen3.8 Max |
|---|---|---|
| LMArena Vision | — | 1314 |
| Furniture Assembly | — | 20% |

## Multilingual

- Devstral Small 2505: —
- Qwen3.8 Max: 56.7 (#18)

| Benchmark | Devstral Small 2505 | Qwen3.8 Max |
|---|---|---|
| LMArena Non-English | — | 1472 |
| LMArena Chinese | — | 1538 |
| LMArena French | — | 1503 |
| LMArena German | — | 1483 |
| LMArena Japanese | — | 1467 |
| LMArena Korean | — | 1461 |
| LMArena Russian | — | 1481 |
| LMArena Spanish | — | 1492 |

## Instruction Following

- Devstral Small 2505: —
- Qwen3.8 Max: 77.6 (#17)

| Benchmark | Devstral Small 2505 | Qwen3.8 Max |
|---|---|---|
| LMArena Instruction Following | — | 1479 |

## Long Context

- Devstral Small 2505: —
- Qwen3.8 Max: 45.6 (#31)

| Benchmark | Devstral Small 2505 | Qwen3.8 Max |
|---|---|---|
| LMArena Longer Query | — | 1489 |

## Writing & Preference

- Devstral Small 2505: —
- Qwen3.8 Max: 67.1 (#30)

| Benchmark | Devstral Small 2505 | Qwen3.8 Max |
|---|---|---|
| LMArena Text | — | 1483 |
| LMArena Creative Writing | — | 1479 |
| LMArena Multi-Turn | — | 1489 |

## FAQ

### Is Devstral Small 2505 better than Qwen3.8 Max?

Qwen3.8 Max is the stronger model overall, scoring 56.8 to 34.3 on the Noometry Index. Devstral Small 2505 costs 20× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.

### Which is cheaper, Devstral Small 2505 or Qwen3.8 Max?

Devstral Small 2505 is cheaper. It lists at $0.10 per million input tokens and $0.30 per million output tokens; Qwen3.8 Max lists at $2 and $6.

### Is Devstral Small 2505 or Qwen3.8 Max better for coding?

Qwen3.8 Max scores higher on coding benchmarks: 53.5 versus 38.9 in the Noometry coding category.

### Which has the bigger context window?

Qwen3.8 Max does, with 1M tokens against 128K.

### How many benchmarks do Devstral Small 2505 and Qwen3.8 Max share?

2 benchmarks have published results for both models. Devstral Small 2505 has 4 scored results on Noometry and Qwen3.8 Max has 39.