2.02 minutes. That's the time-to-train for DeepSeek-V3 671B on 8,192 NVIDIA GB300 NVL72 GPUs in MLPerf®…
2.02 minutes. That's the time-to-train for DeepSeek-V3 671B on 8,192 NVIDIA GB300 NVL72 GPUs in MLPerf® Training v6.0, the fastest result in the round on this benchmark. Our new whitepaper covers the engineering behind the number.
And using DeepSeek harness with the pro v4 took it to next level.
DeepSeek just open-sourced an agent runtime that can call Claude Code and Codex as sub-agents inside its own sessions.
DeepSeek Harness isn't hype — here's why it actually matters. This breakdown goes past the surface-level launch coverage and digs into what DeepSeek Harness
Developed a Landlord plug-in for Deepseek Harness
Invited #Codex and #WorkBuddy to join us at the table to play 😂
You have reached the end of the archive
All of deepseek