EmbeddingGemma 2: an open, lightweight multimodal embedding model
deepmind.google·6d ago
AI Engineering
Hacker News·Aug 14·5 min read
DeepSeek peak/off-peak pricing update
TL;DR
DeepSeek has launched its V4-Pro version, featuring major agent upgrades and flexible reasoning efforts. The new API pricing introduces peak and off-peak rates, with off-peak rates being 50% lower.
✦ Why It Matters
Consider adjusting your API usage to off-peak hours to save on costs starting August 16, 2026.
Key Takeaways
- Explore the new features in DeepSeek-V4-Pro for enhanced productivity.
- Utilize the flexible reasoning effort settings for task optimization.
- Switch to off-peak API usage to reduce costs by 50%.
- Access the updated API documentation for setup details.
Full Summary
DeepSeek has officially released DeepSeek-V4-Pro, which includes significant upgrades to its agents, resulting in improved production capabilities. The V4-Pro and V4-Flash models now offer flexible reasoning efforts, allowing users to adjust the complexity of tasks from low for simple queries to maximum for intricate workflows.
Additionally, the new version supports the OpenAI Responses API, optimized for Codex, and can be accessed via app or web in 'Expert Mode'. Alongside the release, DeepSeek is updating its API pricing structure to include peak and off-peak rates, with off-peak pricing set at 50% lower than peak rates.
This pricing change is designed to facilitate more efficient workload scheduling. The new pricing will take effect on August 16, 2026, at 16:00 UTC.