DeepSeek-V4-Pro Launches on DeepInfra with 1.6T MoE and 1M Context
gharik · x · 2026-08-15
DeepSeek-V4-Pro-0813 is now available on DeepInfra. The model features a 1.6T MoE architecture with 49B active parameters and a 1M context window. Hosted in the US on NVIDIA Blackwell GPUs, it offers priority and flex service tiers along with prompt cache retention. Pricing is set at $1.30 per million input tokens, $2.60 for output, and $0.10 for cached input.
Related event: DeepSeek V4 Series Debuts on Multiple Platforms with 1M Context(6 posts)→
More from Infra
- New hardware day: local LLM deployment the moment it arrives — lxfater · 2026-10-16
- Together AI's score centering fixes RL training collapse under quantized inference — nrehiew_ · 2026-10-16
- Inference is splitting prefill and decode across specialized chips, not just H100s — AccBalanced · 2026-10-16
- Rebasely launches in beta: Git hosting built for humans and coding agents — handotdev · 2026-10-16
- Drama3 TTS lands in audio.cpp: up to 5x real-time on RTX 5090, <8GB VRAM mode too — Acceptable-Cycle4645 · 2026-10-16
- Three CoreWeave co-founders have sold $3.3B in shares since IPO amid credit worries — SumitGup · 2026-10-16