Back
AI summary
Written by AI from the official notes. Check them for exact details.The deepseek-reasoner model has been upgraded to DeepSeek-R1-0528, improving reasoning capabilities and reducing hallucinations.
- Enhanced reasoning capabilities with significant benchmark improvements.
- Optimized front-end development for better aesthetics in web pages and games.
- Reduced hallucination issues from the legacy R1 version.
- Added JSON output and function calling support.
Why it matters: Users relying on reasoning tasks and web development should consider upgrading for better performance and reduced errors.
Full release notes8 changes
deepseek-reasoner Model Upgraded to DeepSeek-R1-0528:
- Enhanced Reasoning Capabilities Significant benchmark improvements (Pass@1) AIME 2025: 70.0 → 87.5 (+17.5)
- GPQA: 71.5 → 81.0 (+9.5)
- LCB_v6: 63.5 → 73.3 (+9.8)
- Aider: 57.0 → 71.6 (+14.6)
- Note: Complex reasoning tasks may consume more tokens compared to legacy R1 version.
- Optimized Front-end Development Generated web pages and games now feature improved aesthetics.
- Reduced Hallucinations Significantly suppressed hallucination issues present in legacy R1 version.
- JSON Output & Function Calling Support Function call performance: Tau-bench score: 53.5 (Airline) / 63.9 (Retail)