Up to four models
DeepSeek V4.1 Flash 1017 vs Claude Opus 4.8
On the index (GPT-4 = 100), DeepSeek V4.1 Flash 1017 scores 172 and Claude Opus 4.8 170. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
172156–188
WeightsOpen
ReleasedOct 2026
Price in / out$0.036 / $0.50
Context1.05M
Core tests taken6 of 9
Claude Opus 4.8
170154–186
WeightsClosed
ReleasedMay 2026
Price in / out$5.0 / $25
Context1M
Core tests taken6 of 9
DeepSeek V4.1 Flash 1017 and Claude Opus 4.8 are too close to call (2 points apart, within the uncertainty), ahead on 5 of 6 shared core tests, and costs 66× less.
Test by testthird-party results
Every result as a table
| Test | DeepSeek V4.1 Flash 1017 | Claude Opus 4.8 |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 93.2% | 91.0% |
| SimpleQA Verified | 42.5% | 53.0% |
| Maths | ||
| FrontierMath (tiers 1-3) | 71.2% | 80.0% |
| AIME-style maths (OTIS mock) | 99.4% | 98.3% |
| Chess puzzles | 46.0% | 52.0% |
| FrontierMath (tier 4) | 46.3% | 56.1% |
| Coding | ||
| Terminal-Bench 4.0 | 31.8% | 23.2% |
| Code Migration | 44.2% | 47.3% |
| Vibe Code Bench v1.1 | 87.7% | 82.7% |
| Agents | ||
| APEX-Agents | 69.8% | 48.9% |
| Mystery game puzzles | 49.0% | 36.0% |
| Professional work | ||
| Finance Agent (v2) | 57.0% | 53.9% |
| EMB | 61.8% | 69.4% |
| Legal Research Bench | 45.2% | 43.8% |
| Tax Agent Bench | 44.4% | 30.0% |
| Novel problems | ||
| ARC-AGI-2 | 78.1% | 72.1% |
| ARC-AGI-1 | 94.5% | 92.5% |
| ARC-AGI-3 | 0.2% | 1.5% |
Tests at least two of these models have taken. Bold rows are the core tests.