Chinese AI firm DeepSeek assembles a Beijing-based team to develop a native agentic coding tool at a fraction of Claude Opus 4.7’s cost.
DeepSeek is forming a Beijing-based team to build a native agentic coding tool, Code Harness, directly competing with Anthropic’s Claude Code and OpenAI’s Codex. The move aims to control the full agentic stack rather than just supplying the underlying model.
DeepSeek V4 already integrates with Claude Code, but its V4 Flash model offers a structural price advantage at $0.14 per million input tokens, up to 100 times cheaper than Claude Opus 4.7. The initiative began as a side project but is now a strategic priority for the firm.
The new team will operate from Beijing, reflecting the city’s central role in China’s AI ambitions and close government oversight. Job postings specify roles for a product manager and R&D engineer, both requiring on-site presence in the capital.