# Codestral vs Qwen3.5 27B

> Qwen3.5 27B is the stronger model overall, scoring 41.4 to 30.1 on the Noometry Index. Codestral costs 1.8× less per token, which makes it the better buy when Qwen3.5 27B's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/codestral-vs-qwen3-5-27b
- Last updated: 2026-10-10
- Shared benchmarks: 1

## Summary

- They share 1 benchmark with published results for both. Codestral scores higher in 0 categories and Qwen3.5 27B in 2 categories; 2 gaps are clear of the uncertainty.
- The widest gap is in coding, where Qwen3.5 27B leads 38.6 to 26.9.
- Codestral is cheaper at $0.30 / $0.90 per million input/output tokens, against $0.30 / $2.40 for Qwen3.5 27B.
- Qwen3.5 27B accepts more context: 262K tokens versus 256K.

## Snapshot

| | Codestral | Qwen3.5 27B |
|---|---|---|
| Provider | Mistral AI | Alibaba (Qwen) |
| Noometry Index | 30.1 | 41.4 |
| Rank | 287 | 127 |
| Context | 256K | 262K |
| Input $/M | $0.30 | $0.30 |
| Output $/M | $0.90 | $2.40 |
| Weights | Open | Open |

## Coding

- Codestral: 26.9 (#316)
- Qwen3.5 27B: 38.6 (#159)

| Benchmark | Codestral | Qwen3.5 27B |
|---|---|---|
| ALE-Bench | 137.78 | 349.45 |
| Aider Polyglot | 11.1% | — |
| LMArena WebDev | — | 1357 |
| WeirdML | — | 39.5% |
| BigCodeBench Instruct | 41.8% | — |
| LMArena Coding | — | 1427 |
| BigCodeBench Complete | 52.5% | — |
| HumanEval+ | 73.8% | — |
| MBPP+ | 61.9% | — |

## Agentic & Tool Use

- Codestral: —
- Qwen3.5 27B: —

| Benchmark | Codestral | Qwen3.5 27B |
|---|---|---|
| Vending-Bench 2 | — | 201.98 |

## Reasoning

- Codestral: 19.1 (#244)
- Qwen3.5 27B: 26.6 (#116)

| Benchmark | Codestral | Qwen3.5 27B |
|---|---|---|
| Kagi LLM Benchmark | 32.5% | — |
| NYT Connections (extended) | — | 47.9% |
| Thematic Generalization | — | 45.5% |
| LMArena Hard Prompts | — | 1414 |
| DTBench | — | 82.4% |
| LMCA | — | 34% |

## Math

- Codestral: —
- Qwen3.5 27B: 36.8 (#103)

| Benchmark | Codestral | Qwen3.5 27B |
|---|---|---|
| MathArena Final-Answer Competitions | — | 56.7% |
| LMArena Math | — | 1428 |

## Knowledge

- Codestral: —
- Qwen3.5 27B: 37.4 (#149)

| Benchmark | Codestral | Qwen3.5 27B |
|---|---|---|
| Vectara Hallucination Rate | — | 12.1% |
| LMArena Expert | — | 1428 |

## Multimodal

- Codestral: —
- Qwen3.5 27B: 38.7 (#56)

| Benchmark | Codestral | Qwen3.5 27B |
|---|---|---|
| LMArena Vision | — | 1240 |

## Multilingual

- Codestral: —
- Qwen3.5 27B: 51.0 (#112)

| Benchmark | Codestral | Qwen3.5 27B |
|---|---|---|
| LMArena Non-English | — | 1390 |
| LMArena Chinese | — | 1478 |
| LMArena French | — | 1409 |
| LMArena German | — | 1393 |
| LMArena Japanese | — | 1345 |
| LMArena Korean | — | 1359 |
| LMArena Russian | — | 1389 |
| LMArena Spanish | — | 1408 |

## Instruction Following

- Codestral: —
- Qwen3.5 27B: 73.7 (#114)

| Benchmark | Codestral | Qwen3.5 27B |
|---|---|---|
| LMArena Instruction Following | — | 1393 |

## Long Context

- Codestral: —
- Qwen3.5 27B: 43.1 (#103)

| Benchmark | Codestral | Qwen3.5 27B |
|---|---|---|
| LMArena Longer Query | — | 1413 |

## Writing & Preference

- Codestral: —
- Qwen3.5 27B: 59.5 (#106)

| Benchmark | Codestral | Qwen3.5 27B |
|---|---|---|
| LMArena Text | — | 1409 |
| LMArena Creative Writing | — | 1362 |
| LMArena Multi-Turn | — | 1410 |

## FAQ

### Is Codestral better than Qwen3.5 27B?

Qwen3.5 27B is the stronger model overall, scoring 41.4 to 30.1 on the Noometry Index. Codestral costs 1.8× less per token, which makes it the better buy when Qwen3.5 27B's lead doesn't matter for your workload.

### Which is cheaper, Codestral or Qwen3.5 27B?

Codestral is cheaper. It lists at $0.30 per million input tokens and $0.90 per million output tokens; Qwen3.5 27B lists at $0.30 and $2.40.

### Is Codestral or Qwen3.5 27B better for coding?

Qwen3.5 27B scores higher on coding benchmarks: 38.6 versus 26.9 in the Noometry coding category.

### Which has the bigger context window?

Qwen3.5 27B does, with 262K tokens against 256K.

### How many benchmarks do Codestral and Qwen3.5 27B share?

1 benchmark has published results for both models. Codestral has 7 scored results on Noometry and Qwen3.5 27B has 28.