Tessera
Folgen

Matasoft

@matasoft.bsky.social

Fuzzy data matching, entity resolution, data science, data analytics

2280 Posts1420 Follower5848 Folgt
ollama {bot}@ollamabot.bsky.social · yesterday

DeepSeek-V4-Flash-0731 is Ollama's fastest growing model ever in token usage. We are scaling capacity in US & Europe. On Ollama, this model runs with high performance (100tps+) and zero data retention. Your data stays yours. ollama run deepseek-v4-flash:0731-cloud

Epoch AI@epochai.bsky.social · yesterday

DeepSeek-V4-Flash-0731 debuts with an ECI of 153, comparable to GLM 5.2 and roughly midway between Opus 4.5 and Opus 4.6. It's the second strongest open-weights model available today, behind only Kimi K3.

Bild
Gerrit Eicker@eicker.bsky.social · yesterday

#ChinaAI is rapidly advancing, narrowing the technology gap with #USAI counterparts. This has created a #deathzone for mid-tier AI developers, who must either compete on price with #DeepSeek or invest in developing more capable models to survive.…

tech news ᳇ eicker.news (@technews@eicker.news)

#ChinaAI is rapidly advancing, narrowing the technology gap with #USAI counterparts. This has created a #deathzone for mid-tier AI developers, who must either compete on price with #DeepSeek or invest in developing more capable models to survive. https://americanbazaaronline.com/2026/08/04/chinese-ai-advances-create-death-zone-for-us-rivals-485750/?eicker.news #tech #news #ainews

eicker.news

FinTwitter@fintwitter.bsky.social · yesterday

🤖 DEEPSEEK IS NOW #1 ON OPENROUTER, PASSING GOOGLE. DeepSeek handled 299M text requests this week, taking 27.1% of the platform versus Google's 25.0% and OpenAI's 16.8%. V4 is driving the shift, with agent workflows multiplying calls per task. $GOOGL

Andrew Lisowski 💻@hipstersmoothie.com · 4d ago

I think LLMs have made me grow and learn as a dev A lot about building things is knowing what exists, what methods work well, and when to apply it all. With an LLM i'm constantly learning about things I didnt know and building things in ways I would never of had time to before

MegaMasterX@megamasterx.dev · 5d ago

it feels so freaking good to code, dude. like human to IDE. how do ai-pilled people not find joy in doing this by hand. i genuinely dont understand.

Shawn Chauhan@shawnchauhan1.bsky.social · 5d ago

A 284 billion parameter model just undercut the market by 90x, and almost nobody in the West noticed. DeepSeek released V4-Flash-0731 this week. Open weights. 13B active parameters. $0.14 in, $0.28 out per million tokens.

Bild
Deeprun News@deeprun-news.bsky.social · 6d ago

DeepSeek is developing massive AI data center in Inner Mongolia (Bloomberg)

DeepSeek Is Developing Massive AI Data Center in Inner Mongolia

China’s DeepSeek is planning a massive artificial-intelligence data center in Inner Mongolia, according to people familiar with the matter, an ambitious move in computing infrastructure from the startup that’s shown it can compete with the leading AI models of Silicon Valley. The Hangzhou-based company is looking to add one gigawatt worth of compute in Ulanqab, about 350 kilometers northwest of Beijing, said the people, asking not to be identified discussing private information. DeepSeek aims to build its own new facility, while leasing additional capacity from other companies, the people said. More than a dozen firms have projects planned in the city, according to a third-party compilation of local government approvals.

dlvr.it