Archive — CN — technode
DeepSeek puts V4-Flash API into public beta
DeepSeek has launched the public beta of its V4-Flash API, enhancing agent tasks with significant benchmark scores. The model retains its structure from the preview version while boasting improved performance. Additional features include support for the Responses API, although the V4-Pro API remains unchanged.
Key points
- DeepSeek's V4-Flash API entered public beta.
- The upgrade primarily enhances agent tasks.
- V4-Flash model scored 82.7 on Terminal Bench 2.1.
- Scored 54.4 on DeepSWE among other metrics.
- Added support for the Responses API.
- Model structure and size unchanged from the preview version.
- The update is exclusive to V4-Flash API; V4-Pro remains unaffected.
Why it matters
- The public beta aligns with the growing trend in APIs focused on improving agent and automation tasks.
- Reflects the current emphasis on benchmarking in AI tools as organizations prioritize performance metrics.
- Similar innovations are crucial in the IT landscape of 2025, where user interaction and API efficiency will dominate.
The release of the V4-Flash API aligns with industry trends aimed at enhancing automation, showcasing DeepSeek's commitment to performance improvement and adaptability.