Model comparison
Claude Sonnet 4.6 vs Devstral Small 2505
Claude Sonnet 4.6 is the stronger model overall, scoring 50.1 to 33.9 on the Noometry Index. Devstral Small 2505 costs 40× less per token, which makes it the better buy when Claude Sonnet 4.6's lead doesn't matter for your workload.
Last verified . 2 shared benchmarks.
Mistral AISummary
- They share 2 benchmarks with published results for both. Claude Sonnet 4.6 scores higher in 2 categories and Devstral Small 2505 in 0 categories; 2 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Claude Sonnet 4.6 leads 46.1 to 19.1.
- The biggest single-benchmark swing is SciCode: 46.8% for Claude Sonnet 4.6 and 28.8% for Devstral Small 2505.
- Devstral Small 2505 is cheaper at $0.10 / $0.30 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.6.
- Claude Sonnet 4.6 accepts more context: 1M tokens versus 128K.
- Devstral Small 2505 has downloadable open weights; the other is API-only.
Side by side
| Claude Sonnet 4.6 | Devstral Small 2505 | |
|---|---|---|
| Provider | Anthropic | Mistral AI |
| Noometry Index | 50.1 | 33.9 |
| Released | 2026-02-17 | 2025-05-07 |
| Weights | Proprietary | Open |
| Context window | 1M | 128K |
| Max output | 128K | 128K |
| Input $ / M tokens | $3 | $0.10 |
| Output $ / M tokens | $15 | $0.30 |
| Results tracked | 57 | 4 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 46.0 (#62), Devstral Small 2505: 38.6 (#160)
| Benchmark | Claude Sonnet 4.6 | Devstral Small 2505 |
|---|---|---|
| SciCode | 46.8% | 28.8% |
| SWE-bench Verified | 75.2% | — |
| DeepSWE | 29.9% | — |
| FrontierCode | 24.3% | — |
| SWE-bench Verified (bash only) | — | 56.4% |
| LMArena WebDev | 1521 | — |
| WeirdML | 66.1% | — |
| LMArena Coding | 1504 | — |
| ALE-Bench | 1,327 | — |
Agentic & Tool Use Not comparable
Claude Sonnet 4.6: 39.1 (#26), Devstral Small 2505: —
| Benchmark | Claude Sonnet 4.6 | Devstral Small 2505 |
|---|---|---|
| Terminal-Bench | 53.4% | — |
| APEX-Agents | 43% | — |
| OSWorld 2.0 | 9.3% | — |
| DeepResearch Bench | 54.9% | — |
| OSWorld | 72.1% | — |
| ExploitBench | 23.6% | — |
| GBAEval | 48.8% | — |
| GDP.pdf | 18% | — |
| LMArena Search | 1221 | — |
| Vending-Bench 2 | 7,204 | — |
Reasoning Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 46.1 (#40), Devstral Small 2505: 19.1 (#243)
| Benchmark | Claude Sonnet 4.6 | Devstral Small 2505 |
|---|---|---|
| CritPt | 3.1% | 0% |
| ARC-AGI-2 | 60.4% | — |
| Kagi LLM Benchmark | — | 37.7% |
| NYT Connections (extended) | 80.9% | — |
| ARC-AGI-1 | 86.5% | — |
| Chess Puzzles | 13% | — |
| Thematic Generalization | 76.3% | — |
| LMArena Hard Prompts | 1485 | — |
| Mystery Game Puzzles | 16% | — |
| DTBench | 89.9% | — |
| LMCA | 46.5% | — |
| Epoch Capabilities Index | 152.25 | — |
| ForecastBench | 62 | — |
Math Not comparable
Claude Sonnet 4.6: 51.9 (#41), Devstral Small 2505: —
| Benchmark | Claude Sonnet 4.6 | Devstral Small 2505 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 85.8% | — |
| ProofBench | 45% | — |
| LMArena Math | 1463 | — |
| FrontierMath (Feb 2025 set) | 32.4% | — |
| FrontierMath Tier 4 (v1) | 8.3% | — |
Knowledge Not comparable
Claude Sonnet 4.6: 51.6 (#60), Devstral Small 2505: —
| Benchmark | Claude Sonnet 4.6 | Devstral Small 2505 |
|---|---|---|
| GPQA Diamond | 87.4% | — |
| SimpleQA Verified | 35.5% | — |
| Vectara Hallucination Rate | 10.6% | — |
| LMArena Expert | 1500 | — |
Multimodal Not comparable
Claude Sonnet 4.6: 38.0 (#61), Devstral Small 2505: —
| Benchmark | Claude Sonnet 4.6 | Devstral Small 2505 |
|---|---|---|
| LMArena Vision | 1283 | — |
| Blueprint-Bench 2 | 6.7% | — |
| LMArena Document | 1482 | — |
Multilingual Not comparable
Claude Sonnet 4.6: 54.6 (#39), Devstral Small 2505: —
| Benchmark | Claude Sonnet 4.6 | Devstral Small 2505 |
|---|---|---|
| LMArena Non-English | 1440 | — |
| LMArena Chinese | 1491 | — |
| LMArena French | 1465 | — |
| LMArena German | 1428 | — |
| LMArena Japanese | 1419 | — |
| LMArena Korean | 1411 | — |
| LMArena Russian | 1440 | — |
| LMArena Spanish | 1465 | — |
Instruction Following Not comparable
Claude Sonnet 4.6: 77.6 (#23), Devstral Small 2505: —
| Benchmark | Claude Sonnet 4.6 | Devstral Small 2505 |
|---|---|---|
| LMArena Instruction Following | 1476 | — |
Long Context Not comparable
Claude Sonnet 4.6: 45.3 (#40), Devstral Small 2505: —
| Benchmark | Claude Sonnet 4.6 | Devstral Small 2505 |
|---|---|---|
| LMArena Longer Query | 1479 | — |
Writing & Preference Not comparable
Claude Sonnet 4.6: 70.4 (#22), Devstral Small 2505: —
| Benchmark | Claude Sonnet 4.6 | Devstral Small 2505 |
|---|---|---|
| LMArena Text | 1458 | — |
| LMArena Creative Writing | 1435 | — |
| EQ-Bench Creative Writing | 1810 | — |
| EQ-Bench 4 | 1207 | — |
| LMArena Multi-Turn | 1464 | — |
Frequently asked questions
Is Claude Sonnet 4.6 better than Devstral Small 2505?
Claude Sonnet 4.6 is the stronger model overall, scoring 50.1 to 33.9 on the Noometry Index. Devstral Small 2505 costs 40× less per token, which makes it the better buy when Claude Sonnet 4.6's lead doesn't matter for your workload.
Which is cheaper, Claude Sonnet 4.6 or Devstral Small 2505?
Devstral Small 2505 is cheaper. It lists at $0.10 per million input tokens and $0.30 per million output tokens; Claude Sonnet 4.6 lists at $3 and $15.
Is Claude Sonnet 4.6 or Devstral Small 2505 better for coding?
Claude Sonnet 4.6 scores higher on coding benchmarks: 46.0 versus 38.6 in the Noometry coding category.
Which has the bigger context window?
Claude Sonnet 4.6 does, with 1M tokens against 128K.
How many benchmarks do Claude Sonnet 4.6 and Devstral Small 2505 share?
2 benchmarks have published results for both models. Claude Sonnet 4.6 has 57 scored results on Noometry and Devstral Small 2505 has 4.