DeepSeek-V4-Pro launches: 1.6T-param MoE cuts inference FLOPs to 27% of V3.2
AccBalanced · x · 2026-08-14
DeepSeek officially released its flagship model DeepSeek-V4-Pro, a 1.6T-total / 49B-active MoE with hybrid CSA+HCA attention and manifold-constrained hyper-connections, achieving 27% of V3.2's per-token inference FLOPs and 10% of KV cache at 1M context. Pre-trained on 32T+ tokens with Muon optimizer, it supports FP4+FP8 mixed precision and native OpenAI Responses API, optimized for Codex. vLLM support is ready with no config rebuild. DeepSeek also open-sourced an agent harness compatible with any OpenAI-compatible endpoint.
Related event: DeepSeek Releases V4 Pro(5 posts)→
More from Infra
- First AA-AgentPerf results: NVIDIA Vera Rubin NVL72 delivers up to 5.7x throughput per MW vs GB300 — ArtificialAnlys · 2026-10-16
- Vera Rubin NVL72 first benchmarks: up to 5.7x throughput per MW vs GB300 — ArtificialAnlys · 2026-10-16
- NVIDIA Vera Rubin NVL72 hits up to 5.7x GB300 throughput per MW on agent benchmark — ArtificialAnlys · 2026-10-16
- 14 open-source LLM observability tools to know in 2026, from Langfuse to Laminar — TheTuringPost · 2026-10-16
- Local AI is a routing problem, not a privacy feature: where each task should run — bigdata · 2026-10-16
- Moving agent work from GPUs to CPUs for durable, predictable actions — cramforce · 2026-10-16