DeepSeek claims that its newly generally available model, DeepSeek V4 Flash, achieves performance parity with Anthropic's Sonnet 5 and xAI's Grok 4.5 on the DeepSWE benchmark.

  • DeepSeek V4 Flash is now generally available.
  • The model ranks identically to Sonnet 5 and Grok 4.5 on DeepSWE.
  • These results are claimed by DeepSeek but have not yet been verified by the DeepSWE team.

The announcement highlights DeepSeek's competitive standing in coding benchmarks, though independent verification is pending.