Model comparison
Claude Opus 4.5 vs Devstral Medium
Claude Opus 4.5 has enough public results to be ranked (#47); Devstral Medium does not yet, so treat this comparison as directional.
Last verified . 2 shared benchmarks.
Mistral AISummary
- They share 2 benchmarks with published results for both. Claude Opus 4.5 scores higher in 2 categories and Devstral Medium in 0 categories; 2 gaps are clear of the uncertainty.
- The widest gap is in coding, where Claude Opus 4.5 leads 54.8 to 24.3.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 80.2% for Claude Opus 4.5 and 37.7% for Devstral Medium.
- Devstral Medium is cheaper at $0.40 / $2 per million input/output tokens, against $5 / $25 for Claude Opus 4.5.
- Devstral Medium accepts more context: 262K tokens versus 200K.
- Devstral Medium has downloadable open weights; the other is API-only.
Side by side
| Claude Opus 4.5 | Devstral Medium | |
|---|---|---|
| Provider | Anthropic | Mistral AI |
| Noometry Index | 50.5 | 30.1 |
| Released | 2025-11-01 | 2025-07-10 |
| Weights | Proprietary | Open |
| Context window | 200K | 262K |
| Max output | 64K | 262K |
| Input $ / M tokens | $5 | $0.40 |
| Output $ / M tokens | $25 | $2 |
| Results tracked | 69 | 2 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Opus 4.5 leads
Claude Opus 4.5: 54.8 (#27), Devstral Medium: 24.3
| Benchmark | Claude Opus 4.5 | Devstral Medium |
|---|---|---|
| LMArena WebDev | 1494 | 1076 |
| SWE-bench Verified | 76.7% | — |
| SWE-bench Verified (bash only) | 76.8% | — |
| SWE-bench Multilingual | 70.7% | — |
| GSO | 26.5% | — |
| WeirdML | 63.7% | — |
| LMArena Coding | 1504 | — |
| ALE-Bench | 1,025 | — |
| AlgoTune | 1.77 | — |
Agentic & Tool Use Not comparable
Claude Opus 4.5: 47.3 (#12), Devstral Medium: —
| Benchmark | Claude Opus 4.5 | Devstral Medium |
|---|---|---|
| Terminal-Bench | 63.1% | — |
| Berkeley Function Calling Leaderboard | 77.5% | — |
| GDPval | 45.5% | — |
| Remote Labor Index | 3.8% | — |
| τ²-bench Airline | 84% | — |
| τ²-bench Banking | 24.7% | — |
| τ²-bench Retail | 79.6% | — |
| τ²-bench Telecom | 92.3% | — |
| Cybench | 82% | — |
| DeepResearch Bench | 54.8% | — |
| OSWorld | 66.3% | — |
| BALROG | 43.5% | — |
| LMArena Search | 1180 | — |
| METR Time Horizons | 75% | — |
| Vending-Bench 2 | 4,967 | — |
Reasoning Claude Opus 4.5 leads
Claude Opus 4.5: 42.6 (#51), Devstral Medium: 21.1
| Benchmark | Claude Opus 4.5 | Devstral Medium |
|---|---|---|
| Kagi LLM Benchmark | 80.2% | 37.7% |
| ARC-AGI-2 | 37.6% | — |
| SimpleBench | 62% | — |
| NYT Connections (extended) | 52.5% | — |
| ARC-AGI-1 | 80% | — |
| Chess Puzzles | 12% | — |
| EnigmaEval | 11.9% | — |
| EBR-Bench | 14.3% | — |
| LMArena Hard Prompts | 1476 | — |
| Mystery Game Puzzles | 22% | — |
| DTBench | 89.9% | — |
| LMCA | 44.5% | — |
| Epoch Capabilities Index | 150.09 | — |
| ForecastBench | 60.7 | — |
Math Not comparable
Claude Opus 4.5: 38.6 (#132), Devstral Medium: —
| Benchmark | Claude Opus 4.5 | Devstral Medium |
|---|---|---|
| FrontierMath (Tiers 1-3) | 34.4% | — |
| FrontierMath Tier 4 | 4.9% | — |
| OTIS Mock AIME 2024-2025 | 86.1% | — |
| ProofBench | 36% | — |
| LMArena Math | 1463 | — |
| FrontierMath (Feb 2025 set) | 20.7% | — |
| FrontierMath Tier 4 (v1) | 4.2% | — |
Knowledge Not comparable
Claude Opus 4.5: 56.5 (#44), Devstral Medium: —
| Benchmark | Claude Opus 4.5 | Devstral Medium |
|---|---|---|
| GPQA Diamond | 86% | — |
| Humanity's Last Exam | 25.2% | — |
| SimpleQA Verified | 45.7% | — |
| Vectara Hallucination Rate | 10.9% | — |
| LMArena Expert | 1487 | — |
Multimodal Not comparable
Claude Opus 4.5: 31.4 (#107), Devstral Medium: —
| Benchmark | Claude Opus 4.5 | Devstral Medium |
|---|---|---|
| GeoBench | 75% | — |
| VPCT | 40% | — |
| Furniture Assembly | 28.3% | — |
| LMArena Document | 1462 | — |
Multilingual Not comparable
Claude Opus 4.5: 54.3 (#47), Devstral Medium: —
| Benchmark | Claude Opus 4.5 | Devstral Medium |
|---|---|---|
| LMArena Non-English | 1438 | — |
| LMArena Chinese | 1470 | — |
| LMArena French | 1471 | — |
| LMArena German | 1449 | — |
| LMArena Japanese | 1416 | — |
| LMArena Korean | 1424 | — |
| LMArena Russian | 1447 | — |
| LMArena Spanish | 1458 | — |
Instruction Following Not comparable
Claude Opus 4.5: 77.5 (#19), Devstral Medium: —
| Benchmark | Claude Opus 4.5 | Devstral Medium |
|---|---|---|
| LMArena Instruction Following | 1478 | — |
Long Context Not comparable
Claude Opus 4.5: 46.5 (#22), Devstral Medium: —
| Benchmark | Claude Opus 4.5 | Devstral Medium |
|---|---|---|
| CL-bench | 21.1% | — |
| LMArena Longer Query | 1480 | — |
Writing & Preference Not comparable
Claude Opus 4.5: 68.1 (#28), Devstral Medium: —
| Benchmark | Claude Opus 4.5 | Devstral Medium |
|---|---|---|
| LMArena Text | 1451 | — |
| LMArena Creative Writing | 1445 | — |
| EQ-Bench Creative Writing | 1687 | — |
| LMArena Multi-Turn | 1466 | — |
Frequently asked questions
Is Claude Opus 4.5 better than Devstral Medium?
Claude Opus 4.5 has enough public results to be ranked (#47); Devstral Medium does not yet, so treat this comparison as directional.
Which is cheaper, Claude Opus 4.5 or Devstral Medium?
Devstral Medium is cheaper. It lists at $0.40 per million input tokens and $2 per million output tokens; Claude Opus 4.5 lists at $5 and $25.
Is Claude Opus 4.5 or Devstral Medium better for coding?
Claude Opus 4.5 scores higher on coding benchmarks: 54.8 versus 24.3 in the Noometry coding category.
Which has the bigger context window?
Devstral Medium does, with 262K tokens against 200K.
How many benchmarks do Claude Opus 4.5 and Devstral Medium share?
2 benchmarks have published results for both models. Claude Opus 4.5 has 69 scored results on Noometry and Devstral Medium has 2.