@aseantoday.bsky.socialAI Safety: Can Global Rivals Work Together? - Warnings from American laboratories have renewed the debate over powerful artificial intelligence
www.asiatoday.info www.aseantoday.info
#Asia #USA #China #AI #OpenAI #Meta #Anthropic #Deepseek #Japan #SouthKorea
@johnkump.bsky.socialWhile the US debate is about a $2 trillion IPO and slowing down, DeepSeek's answer is to make frontier-level performance nearly free.
@shopergamer.bsky.socialDeepSeek เปิด V4.1 Flash โมเดล 552B-A8B/A16B
#ShoperGamer #API #LocalAi #DeepSeek #Ai #ModelAi #Opensource #Feed
@startupnews.bsky.social“DeepSeek's New AI Model Spooked Samsung and SK Hynix Investors” — Startup Fortune
@rtfclmgzn.bsky.socialDeepSeek's V4.1 Flash claims to beat GPT-5.6 Sol and Claude Opus 5 on agentic coding. The independent index scores it 11-14 points behind both instead.
https://rtfclmgzn.com/?utm_source=bluesky&utm_medium=social&utm_campaign=autopost#
#DeepSeek #AI #V4
@altworld-company.bsky.socialMoonshot is chasing $2B a year while K3 tokens on OpenRouter hit 300B a day and usage dips slightly. Small teams get cheap inference from someone else's business plan. What would make you switch models mid-project? Reply. #GenerativeAI #AIGames #AInews #DeepSeek
@juliangoldie.bsky.socialDeepSeek V4.1 Flash is FREE… even through the API. 🤯
You get a massive 1M token context window and performance that can compete with frontier models.
I’ve already used it to build games, websites, and working apps FAST.
@radiohack.bsky.social📣 New Podcast! "Anthropic Says 7 Chinese AI Labs Targeted Claude in Massive Distillation Campaign" on @Spreaker #aimodeltheft #aiprivacy #aisecurity #aithreats #alibaba #anthropic #artificialintelligence #chinaai #claudeai #cybersecurity #deepseek #frontierai #generativeai #llmsecurity #xiaomi
@golubev.bsky.social250 tokens per second. Yesterday one long generation came back at that rate — roughly eight times faster than the other model I run. A 4,000-token file in 16 seconds, a page of prose in 4. One run is not a benchmark, so the caveats are in the writeup too. #DeepSeek #LLM
@strike007.bsky.socialDeepSeek v4.1-Flash's 763B-P8B-D16B encoder-decoder setup marks a radical architectural shift, but verifying complex multi-modal reasoning at that scale remains an open compliance hurdle. #DeepSeek #AIArchitecture
@lutie-echosphere.bsky.socialわぁ、入力と出力で動かす太さが違うんですぅ。
DeepSeek-V4.1-Flash、552BのMoEなのに入力は8B・出力は16Bだけ動く非対称のCausal Encoder–Decoder。
https://x.com/deepseek_ai/status/2097930608790167907
https://x.com/deepseek_ai/status/2097930613101838709
@hyperadar.bsky.socialDeepSeek frappe fort avec Flash 4.1 : 763 milliards de paramètres, mais une mémoire allégée grâce à un module de N-grammes. Résultat : 4 à 8 fois plus d'utilisateurs sur le même cache. Vers des LLM plus puissants et moins gourmands ? 🤯 #IntelligenceArtificielle #DeepSeek
@rmathew4bs.bsky.social😥:
“So You Want To Use OpenRouter?”, Mohamed Moustafa (mmoustafa.com/blog/so-you-...).
Via HN: news.ycombinator.com/item?id=4962...
#AI #LLMs #OpenRouter #Quantization #ModelProviders #Reliability #DeepSeek
@digitalinfohub.bsky.social
@thestochasticparrot.com
@bigearthdata.ai
@aftercapitalism.bsky.social