DeepSeek is a bit special in that things they release are often not knock-your-socks-off in papers, but they *work*. V4-Flash demonstrates this forcefully. It's a no-bullshit…
DeepSeek is a funny company because while they're doing tremendously much for open source, it's not pandering to the "local AI" and "normal people" with 4090s.…
RT Maxime LabonneFun surprise: DeepSeek used my open-perfectblend dataset to train their new DSpark drafterTime to promote it again! It's an open-source reproduction of "The Perfect…
Hello guys, it seems like DeepSeek added a new vision mode to their application. Does this mean, that they will release a new vision model? Edit:…
Article URL: https://github.com/deepseek-ai/DeepSpec/blob/main/DSpark_paper.pdf Comments URL: https://news.ycombinator.com/item?id=48696585 Points: 199 # Comments: 33
I am relatively new here, I have little experience in how long support development takes. I know there are forks. But not merged status means AFAIK…
AI startup Lindy ditched Claude entirely for Deepseek after AI costs exceeded personnel costs. CEO Flo Crivello calls it "a matter of survival for the business."…
DeepSeek Frontier Research GroupHopefully this becomes the most prestigious lab-within-lab in China. Might well happenAiBattle: 2 interesting DeepSeek job postings. Not sure how long they have…
RT Xiuyu LiCode Agent, Generalist Agent 和更贵的 Agent 数据DeepSeek 寻找闪亮发光的你 https://mp.weixin.qq.com/s/LFtjgHJpDt-TPfjD0oUM5w永雏塔菲: deepseek 看来是要猛猛发力了，在现有团队规模基础上至少扩大一倍😲
RT Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)Re Translation of @deepseek_ai's entire recruitment post "At least double all departments" implies more than doubling the whole org, I…
I saw the DS4 repo and the last time I tried it I was just short of 5-10GB of VRAM to fit the model I wanted…
Hello guys, hoping you're doing fine! I was wondering, for users with 4x-8x 6000 PROs (so between 384 and 768GB VRAM), how are bigger models working…
Hey fellow localites, Has anybody used DwarfStar with DeepSeek V4 Flash on 1x DGX Spark yet? What are your thoughts? Based on what I read, with…
btw Zai IPO'ed in Jan at HK$120 a share. when I first met @louszbd nobody really knew anyone using GLM's.now they have beat deepseek with the…
GLM-5.2 should be “DeepSeek moment” for agents. We enter a new world where the top end of agentic capabilities are available in open models.If you care…
RT Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)Dario's answer to Dwarkesh on why he won't buy $300B-1T of compute sure was incomplete. But I guess it would…
Added a DeepSeek Sparse Attention (DSA) from-scratch implementation to my LLMs-from-scratch repo thanks to an awesome new reader contrib. With motivation, overview, and GPT-style model reference…
An eventful month with one flagship release after another
New article: a visual tour of recent LLM architecture advances, from Gemma 4 to DeepSeek V4.I focus on long-context efficiency tweaks like KV sharing, per-layer embeddings,…
First week of Musk v. Altman, OpenAI ends Microsoft legal peril over its $50B Amazon deal, DeepSeek previews new AI model that ‘closes the gap’ with…
Our 243rd episode with a summary and discussion of last week’s big AI news!
The Batch AI News and Insights: I’ve been hearing from people at all levels of seniority about a feeling of job insecurity.
Understanding How DeepSeek's Flagship Open-Weight Models Evolved
Merge pull request #6 from hnyls2002/fine-tune-sglang-command Finetune sglang launching command
DeepSeek shocked the field in late 2024 with DeepSeek-V3, then again with R1, a reasoning model trained at a fraction of the budget Western labs spend. Current lines: V3.2-Exp, R1, V3. The most-talked-about Chinese AI lab of the cycle.
Owner: DeepSeek. We have 280 stories indexed for this model, auto-tagged from titles across every tracked source — official announcements, papers, GitHub release notes, and third-party press. The CTA on each card links to the original; the official site is www.deepseek.com.
Related text models: GPT, Claude, Gemini, Gemma, Llama, Mistral, Grok, Qwen.