AITOPAI热报
ModelsMulti-sourceSelected78°

DeepSeek 552B-parameter MoE model

官方账号深度求索 DeepSeek@deepseek_ai
Post· 3 sources
View the post
Why we picked it

This is a significant upgrade for DeepSeek's flagship model, showing improved performance with a more efficient architecture.

DeepSeek AI announced a new 552B-parameter model with an asymmetric architecture. The model uses 8B active parameters for input and 16B for output. It achieved benchmark results ahead of flagship models like DeepSeek-V4-Pro.

Original · 深度求索 DeepSeek

🧠 Asymmetric architecture. More intelligence, less cost. 🔹 552B-parameter MoE. 🔹 New Causal Encoder–Decoder architecture: just 8B active parameters for input, 16B for output. 🔹 New pre-training methods + larger-scale RL post-training deliver benchmark results ahead of flagship models, including DeepSeek-V4-Pro. 2/6 💬 30 🔄 113 ❤️ 1469 👀 146670 📊 187 ⚡

#DeepSeek#MoE#ai-models#large language model

Keep reading

Recent stories with the same tag or section
ProductsUS vs ChinaSelected07:26
vLLM sets best DeepSeek-V4.1-Flash result on SemiAnalysis AgentX

Matching DeepInfer's performance while beating it on a third-party agentic benchmark suggests vLLM is now a credible production backend for high-throughput DeepSeek serving, not just a research engine. Engineers running agent-heavy inference stacks can use the reproducible blog recipe to benchmark their own deployments against the 954K tokens/s figure. DeepSeek's off-peak pricing makes the revenue-per-GPU-hour number a useful cost baseline for capacity planning.

#vLLM#DeepSeek#AgentX#SemiAnalysis
官方账号vLLM@vllm_projectSource
ModelsBuildersSelected02:02
vLLM Matches DeepInfer on DeepSeek-V4.1-Flash Decode

An open-source inference stack catching up with a frontier lab's in-house engine means teams can serve DeepSeek-V4.1-Flash without relying on the vendor's own tooling. Engineers running self-hosted deployments in Singapore or ASEAN data centres can benchmark vLLM at DEP32 as a direct alternative to DeepInfer's EP64DP64 setup, avoiding engine lock-in while keeping comparable throughput at large batch sizes.

#vLLM#DeepSeek#DeepInfer#inference
官方账号vLLM@vllm_projectSource