Say

On-Premise Business AI Center After my posts on the 2-GPU and 4-GPU builds, people reached out asking how to build an 8-GPU box for their businesses. Why? - Protect their IP - Protect customer data - Save on inference costs - Train their own models Here's how to build one: 🧵

Today we’re launching the OpenAI Deployment Company to help businesses build and deploy AI. It's majority-owned and controlled by OpenAI. It brings together 19 leading investment firms, consultancies, and system integrators to help organizations deploy frontier AI to production for business impact. openai.com/index/openai-l…

New open-source image model:HiDream-O1-Image (8B), including both the undistilled and distilled Dev variants, together with the Reasoning-Driven Prompt Agent. As I said, removing VAE is a trend. Two versions:dev 28 steps of inference, and standard version uses 50 steps.

Tencent HY just released Hy3 preview 👉open source. 295B total, 21B active, 256K context. Hybrid fast-slow thinking MoE. 🚀 First model after a full rebuild of pretraining and RL infrastructure. Biggest gains in coding and agentic tasks. 🛠️ Agent: drives up to 495-step complex workflows in production (docs, data analysis, MCP tool chains) ⚡ Inference: TTFT -54%, end-to-end latency -47%, success rate 99.99%+ on CodeBuddy/WorkBuddy 🎯 Strong on SWE-bench Verified, Terminal-Bench 2.0, BrowseComp, WideSearch — competitive across coding and search agent benchmarks ✅ OpenClaw / OpenCode / KiloCode compatible. vLLM + SGLang supported. 🤖 modelscope.cn/models/Tencent… 💻 github.com/Tencent-Hunyua…

ERNIE 5.1 is here 🚀 ERNIE 5.1 significantly reduces pretraining cost while compressing total parameters to ~1/3 and activated parameters to ~1/2 — using only ~6% of the pretraining cost compared to models at similar scale, while achieving leading performance in its class. 💡Key highlights: 1/ Strong agentic performance approaching leading frontier models. ERNIE 5.1 surpasses DeepSeek-V4-Pro on both τ3-bench and SpreadsheetBench-Verified. 2/ Strong world knowledge and creative writing capabilities, with GPQA and MMLU-Pro performance approaching leading closed-source models, and creative writing ability nearing Gemini 3.1 Pro. 3/ Frontier-level reasoning performance. ERNIE 5.1 scores 99.6 on the challenging AIME26 benchmark with tools, second only to Gemini 3.1 Pro. 4/ Deep search capability. On May 9, ERNIE 5.1 ranked #4 globally and #1 among Chinese models on the Arena Search leaderboard with a score of 1223. ERNIE 5.1 is now available on ERNIE and the Baidu AI Studio Model Playground: 👉ernie.baidu.com 👉aistudio.baidu.com 👉ernie.baidu.com/blog

Today we're releasing ZAYA1-8B, a reasoning MoE trained on @AMD and optimized for intelligence density. With <1B active params, it outperforms open-weight models many times its size on math and reasoning, closing in on DeepSeek-V3.2 and GPT-5-High with test-time compute. 🧵

Here is a 2nd batch of April architecture drops. What a month! - Ant Ling 2.6 1T - Minimax M2.7 - Xiaomi MiMo V2.5 - Poolside Laguna XS.2 - Tencent Hy3-preview - IBM Granite 4.1

> Produce the best fucking LLM models > make it open weights and open source > Prohibit companies to fire people using AI as an excuse

Qwen 3.6 - 27b dense and MOE models are revolutionary models, hopefully we will see that @Alibaba_Qwen will continue working on them and the performances will break the scales.

Mistral Medium 3.5, a new flagship model in public preview by @MistralAI that merges instruction-following, reasoning, and coding into a single 128B dense model with a 256k context window and configurable reasoning effort. It's a new default model for Mistral Vibe and Le Chat. Released as open weights, under a modified MIT license.

The most powerful real-time visual tool in creative coding also has the steepest learning curve Now your Hermes agent can just run TouchDesigner for you. Video credit: made by @macbethAI, a talented AI artist and avid Hermes user, with the TouchDesigner skill

> be Alexandra Elbakyan > be born in Kazakhstan in 1988 > start coding at 12 > hack your internet provider at 14 > hack MIT Press at 16 to download neuroscience books you can't afford > get a CS degree from Satbayev University > intern in neuroscience at Georgia Tech > speak at Harvard on brain-computer interfaces > notice researchers can't read the papers they need > notice academic publishers charging $30 a paper > notice peer reviewers worked for free > notice editors worked for free > notice universities funded the research with billions of dollars of public money > build Sci-Hub in 2011 > upload nearly every paywalled research paper ever published > give it away for free > get sued by Elsevier > get hit with a $15 million judgment > don't give a flying f*ck > keep Sci-Hub up > get domain after domain seized > register a new one > keep Sci-Hub up > get investigated by the US Department of Justice > don't give a flying f*ck > get accused of working for Russian intelligence > don't give a flying f*ck > have the FBI subpoena your iCloud > get named one of Nature's ten people who mattered in science > get a parasitoid wasp named after you > get a deep-sea snail named after you > get the Electronic Frontier Foundation Award for Access to Scientific Knowledge > become a legend

Just dropped two open-source models: MiMo-V2.5-Pro (Code Agent, 1T total) and MiMo-V2.5 (Multimodal Agent, 310B total). Oh and one more thing — we're giving devs & creators 100T tokens on us. Go build something cool 🛠️ 🎁 100T Free Token Grant for Builders 100t.xiaomimimo.com

Discourage people from studying SE, CS and then see people from other continent drop some papers which totally revolutionise the industry. Your tech is complimentary, without devs, people it's nothing, has no economical value. It should be for people, not against them.

Holy shit 2.9% precision lost on UD-IQ3_XXS, the quant I’m using on all my benchmarks! This is insanely good lads, this makes it suitable for daily use and explains all the strong results I shared here in the last couple days 16GB VRAM BROS WE ARE WINNING TODAY! 🥹

I am watching AI race in the front seat with my unemployed ass.

This proves that AI and human creativity are complementary, not substitutes.

Chinese AI labs will take jobs of many CEOs. Tech CEOs are just greedy, emotional beings need to be replaced.

🚨🇨🇳 BREAKING — DeepSeek V4 Drops NVIDIA Huawei Ascend Chips Cut AI Costs 100x Open Source Alternative Scores 3206 Rating Near Global Frontier