DeepSeek's top-ranked V4 Flash stumbles on real agent tasks as its prices surge

DeepSeek’s V4 Flash has become a developer favorite and leaderboard leader, but Composio found it completed only 53.8% of complex multi-step agent tasks across tools like Gmail, GitHub, Slack, and Google Sheets. Results varied sharply by harness and configuration, suggesting orchestration can matter as much as model capability. DeepSeek is also raising prices for V4 Flash and Pro after rapid adoption.

The findings highlight a widening gap between benchmark performance and real production reliability for AI agents. As pricing rises, enterprises may judge models less on raw intelligence or low cost and more on workflow fit, tool integration, and consistency across deployment stacks.