DeepSeek V4, a new iteration of the DeepSeek language model, has been released. Early testing indicates that the model struggles with generating proper tool calls, often taking a significant amount of time to process before failing. AI
IMPACT Early performance issues with tool calls may indicate areas for improvement in future iterations of DeepSeek models.
RANK_REASON New model release from a frontier lab. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →