WHAT NEW IN X10.28 WED

AI collections from X · 22 collections · 58,925 posts

New todaydeepseek-v4.1IMAGE

Strata 0.2.7: DeepSeek V4.1 Flash local prompt processing up to +65%

Strata 0.2.7 speeds up local DeepSeek-V4.1-Flash-1017 inference with up to 65% faster prompt processing, 47% on RTX 5070 at 4K, and native Strix Halo support reaching 300+ tok/s at 4K–32K.

SuSu_酥酥👅
@NFT_ChenArchived 9 hours agoOriginally posted 2026.10.28
Strata 0.2.7: DeepSeek V4.1 Flash local prompt processing up to +65% (image 1 of 2)
Strata 0.2.7: DeepSeek V4.1 Flash local prompt processing up to +65% (image 2 of 2)
SuSu_酥酥👅 @NFT_ChenOriginally posted 2026.10.28
Open original on X

🚨牛哇!DeepSeek-V4.1-Flash-1017 本地推理迎来重大优化!Strata 0.2.7 带来最高 +65% Prompt 速度,Strix Halo 居然也能畅跑 ! Strata 0.2.7 更新核心提升: 🔹Prompt 处理速度最高 +65%(Strix Halo 4K、Radeon 32K 等) 🔹RTX 5070 4K prompt +47%,8K +33% 
🔹正式支持 Ryzen AI Max(Strix Halo)Windows 原生跑 DeepSeek(4K~32K 最高 300+ tok/s) 
🔹4× R9700 默认 5 experts 模式,暖启动 prompt +40% 以上,质量几乎不变 
🔹支持 512K/1M 上下文 + Vision 默认开启 消费级硬件(12GB 卡 + 大内存)也能更流畅地跑 DeepSeek-V4.1-Flash-1017 了! 详情与下载:https://github.com/Niko1221/Strata/releases/tag/v0.2.7 #DeepSeek #DeepSeekV41 #Strata #本地AI #开源 #LocalLLM #AI推理 #梁文峰

This post quotes another post: “Strata 0.2.6 doubles local DeepSeek V4.1 Flash speed on a 4080 Super”

x.com/nft_chen/status/2115123009661145586Archived 2026.10.28 12:04