API pricing for the model starts at $0.15 per million uncached input tokens and $0.60 per million output tokens during off-peak hours. These rates are lower than the $4 and $20 fees for GPT-5.6 Sol, though GPT-6 Luna remains cheaper at $0.10 and $0.50. The new version is optimized for WorkBuddy and Tencent CodeBuddy and features faster task completion and shorter outputs than previous Flash iterations.
Tuesday, October 20, 2026
Story
DeepSeek V4.1 Flash Outperforms GPT 5.6 Sol in Coding Tests
AI4 days ago
The company has made its latest open-weight AI model available on HuggingFace and through its official API, app and website. This V4.1-Flash-1017 release scores 77.0 on the DeepSWE coding benchmark, surpassing 73.0 for GPT-5.6 Sol and 74.0 for Claude Opus 5. It also outperformed those models on the Automation-Bench with a score of 60.5, compared to 45.8 for Sol and 50.3 for Opus 5, while supporting a 1 million token context window.
Looking for more context or analysis?
Tilt chat provides reliable, tailored insights, alerts, and automations.