We self-host DeepSeek V4 Flash on AWS spot instances (twitter.com via hn)
model roundup
DeepSeek 4
-
https://t.co/syBl6fkTDJ Miguel Salinas@VercantezHow we self-host DeepSeek V4 Flash on AWS spot instances4:18 PM · Jul 23, 20269.7KViews243560
-
[AINews] "Laguna S 2.1 Released: Cheaper than Deepseek v4 Flash, Better than V4 Pro" a quiet day lets us highlight a new neolab win. Reignited distillation wars conversation aside, today was more of the same of previous news cycles, which…
-
Full-parameter post-training of trillion-parameter-scale MoE models introduces substantial system-level challenges for large-scale distributed training, including severe memory pressure, non-overlapped communication overhead, and inefficie…
-
An open-source model on par with DeepSeek v4 has appeared in South Korea (old.reddit.com via hn)
could not extract summary
-
Hey Folks, I'm brand new to Claude Code or even coding with AI Agents beyond asking Gemini / GPT over chat. I tried asking Gemini how to setup Claude code with DeepSeek v4 Pro and it worked, but a 2 messages to Deepseek cost me $0.07.
-
Twitter user investigating potential Fable distillation in DeepSeek V4 Pro (twitter.com via hn)
🧵 DeepSeek appear to have engaged, or be engaging in, a large-scale operation to collect outputs from proprietary models (including Claude Fable 5) for certain requests via their API as part of a distillation effort. After seeing such cla…
-
TL;DR: I ran DeepSeek-V4-Flash through the same 89-task Terminal-Bench 2.1 suite on two very different local setups: an aggressively quantized 80.8 GiB GGUF on one 128 GB M5 Max MacBook; the native mixed FP8/FP4 checkpoint with DSpark spec…
-
Fixing Gnome Snapshot Segfault on NixOS and OpenCode with DeepSeek V4 (grigio.org via hn)
TL;DR: GNOME Snapshot segfaults (SIGSEGV 139) when the camera portal returns a PipeWire file descriptor. The portal steals the connection's fd, disconnects the client, and hands the dup to Snapshot — which then tries to re-register on a de…
-
Every viable coding model right now imo (www.reddit.com via reddit)
Mimo V2.5 Deepseek V4 Flash (Max) Mimo V2.5 Pro Deepseek V4 Pro (Max) Composer 2.5 Grok 4.5 (High) GPT 5.6 Sol (Max) / Fable 5 From cheapest to most expensive and best at it's price range these are the most viable models right now; I think…
-
I couldn't pay for DeepSeek V4 with my card. So I built a proxy (api-hub.cc via hn)
DeepSeek V4 is incredible value. But the official platform won't take my credit card.