“The new DeepSite space on Hugging Face is totally insane for vibe-coders https://x.com/Thom_Wolf/status/1906299093518848103
Welcome to the 🤗 AI Agents Course – Hugging Face Agents Course https://huggingface.co/learn/agents-course/en/unit0/introduction
“TL;DR: we are excited to release a powerful new open-weight language model with reasoning in the coming months, and we want to talk to devs about how to make it maximally useful: https://x.com/sama/status/1906793591944646898
“OpenAI CEO Sam Altman also shared that they will release a new open-weights model in the coming months It will be the first since GPT-2 and made “maximally useful” in line with developer inputs from upcoming events A big move! https://x.com/rowancheung/status/1906942132025225492
“Wow. OpenAI is planning to release their first open-weight model since GPT-2 in the next few months.. https://x.com/LiorOnAI/status/1906817391901986965
“The 21st Century Space Race OpenAI just publicly committed to releasing a powerful new open-weight language model focused on reasoning. This would be their first open release since GPT-2, which was initially deemed too dangerous to open-source at the time. According to CEO Sam https://x.com/scaling01/status/1906834038486110594
DeepSite – a Hugging Face Space by enzostvs https://huggingface.co/spaces/enzostvs/deepsite
“we will not do anything silly like saying that you cant use our open model if your service has more than 700 million monthly active users. we want everyone to use it!” / X https://x.com/sama/status/1906845532758405319
“EasyControl just dropped on Hugging Face Adding Efficient and Flexible Control for Diffusion Transformer Free chatgpt style ghibli image generation with easy control https://x.com/_akhaliq/status/1907087568404894136
Alibaba Preparing for Qwen 3 AI Model Release as Soon as April – Bloomberg https://www.bloomberg.com/news/articles/2025-04-01/alibaba-preparing-for-flagship-ai-model-release-as-soon-as-april
“4. @Alibaba_Qwen’s Qwen2.5-Omni Understands any types of input and introduces a two-part Thinker-Talker system and TMRoPE feature to create responses in both text and natural speech. https://x.com/TheTuringPost/status/1907024900747042825
“Introducing Open Responses – the open-source, self-hosted alternative to @OpenAI’s Responses API! With Open Responses, you get: ✅ Multi-provider support (Claude, GPT-4, local models) ✅ Seamless compatibility with OpenAI’s Agents SDK ✅ Full control over your deployment” / X https://x.com/IshitaJindal17/status/1904952986356592708
“Crazy that GPT-2 is still pulling in over 22M downloads in the last 30 days on @huggingface—plus strong activity from Whisper and CLIP-ViT! https://x.com/fdaudens/status/1906807591537148410
“Alibaba dropped Qwen2.5-Omni-7B, a new multimodal AI —Processes across text, audio, image, and video in real-time —Text and speech outputs —Strong speech understanding, outperforms specialized audio models —Can run on phones, laptops —Open-source https://x.com/rowancheung/status/1905130218433450119
“Now in Enterprise Hub organizations, you can centralize your billing not only for HF usage but also inference through our inference partners. Will prevent some headaches for your finance & accounting teams haha (so feel free to share that with them). https://x.com/ClementDelangue/status/1907462470190375109
“30x throughput gain on the Llama 3.1 405B test versus the H200 NVL8 system. With @nvidia ‘s GB200 NVL72 system — connecting 72 NVIDIA Blackwell GPUs to act as a single, massive GPU ⚙️ Time to First Token and Time per Output Token Production settings care about time to first https://x.com/rohanpaul_ai/status/1907864056054296823
“Github 👨🔧: Olares: An Open-Source Sovereign Cloud OS for Local AI Olares transforms personal hardware into a sovereign cloud OS focused on local AI. Host leading open AI models like LLaMA and Stable Diffusion directly on your terms. Easily deploy AI applications from the Olares https://x.com/rohanpaul_ai/status/1906881449384448232
“Neat! Stop wasting time on incompatible AI models: @huggingface’s new feature shows which GGUF quantizations your hardware can run. Add your specs once, see perfect matches forever. Your MacBook won’t catch fire trying to load a 59GB model again. https://x.com/fdaudens/status/1907177449412063326
“@bongrandp @sama @AIatMeta @MistralAI he did not insult them. he merely said that the license was silly and that he’s not going to do that. this is a good thing.” / X https://x.com/cognitivecompai/status/1906923292034044414
“open weight model coming in the next few months” / X https://x.com/snsf/status/1906946307853562084
“📣 In the coming months, we’ll be releasing a highly capable open language model; our first since GPT-2. As @sama shared, we’re hosting sessions with global devs to gather feedback and engage directly with the community to ensure we get this right.” / X https://x.com/kevinweil/status/1906797119848988822
TrustAIRLab/HateBenchSet · Datasets at Hugging Face https://huggingface.co/datasets/TrustAIRLab/HateBenchSet
“𝗩𝗕𝗲𝗻𝗰𝗵-𝟮.𝟬 is out on Hugging Face a next-gen benchmark for evaluating 𝗶𝗻𝘁𝗿𝗶𝗻𝘀𝗶𝗰 𝗳𝗮𝗶𝘁𝗵𝗳𝘂𝗹𝗻𝗲𝘀𝘀 18 fine-grained dimensions Fully automatic & open-source Human-aligned via large-scale validation https://x.com/_akhaliq/status/1906757376507535736
Kvax: Fast and easy-to-use Flash Attention implementation for JAX https://nebius.com/blog/posts/kvax-open-source-flash-attention-for-jax
[2503.24290] Open-Reasoner-Zero: An Open Source Approach to Scaling Up Reinforcement Learning on the Base Model https://arxiv.org/abs/2503.24290
Databricks Has a Trick That Lets AI Models Improve Themselves | WIRED https://www.wired.com/story/databricks-has-a-trick-that-lets-ai-models-improve-themselves/
“Connect a Model Context Protocol (MCP) server with Databricks https://x.com/adnan_hashmi/status/1905826935818572214
“today we’re excited to announce guMCP open-source MCP servers for every app, hosted for free new integrations coming daily to democratize how LLMs get access to data https://x.com/gumloop_ai/status/1905709980327739888
“I built an MCP server for WhatsApp It connects to your personal WhatsApp account You can search your messages, contacts and send messages It’s fully open-source, self-hosted, and doesn’t rely on third-party APIs https://x.com/LukeHarries_/status/1905986562388635913
“Timeline of MCP from my pov Nov: We open source it with a lowkey blog post, I post an announcement tweet. It gets way more attention than I expected given how early the spec was at the time. We realize there’s a lot of demand for some standards here. Dec: We host two MCP” / X https://x.com/alexalbert__/status/1907885414557618406
“📢 I’m excited to share MCP Tools – an open-source CLI I built to solve my own struggles with testing Model Context Protocol servers. If you’re working with MCPs, this might make your life easier too. Simple and interactive. 👉 https://x.com/fkadev/status/1904790444103311644
“🚨BREAKING: DeepSeek-V3-0324🐳 just ranked #5 on the Arena leaderboard – surpassing DeepSeek-R1 and every other open model! Highlights: – #1 open model (MIT license) – 2x cheaper than DeepSeek-R1 – Top-5 across ALL categories – Significant jump over the previous DeepSeek-v3 https://x.com/lmarena_ai/status/1906739061236334744
“🚨 Narrative Violation—DeepSeek V3 is a competitive, but NOT top model. SEAL leaderboards have been updated with DeepSeek V3 (Mar 2025). – 8th on Humanity’s Last Exam (text-only). – 12th on MultiChallenge (multi-turn). View the full rankings: https://x.com/alexandr_wang/status/1907848607635746973
“Want to vibecode with DeepSeek? Just spent 10 minutes with this space and created a full world indicators dashboard – literally just by describing what I wanted! Anyone can now prototype and deploy projects instantly. Think product analytics, market research, or even games – all https://x.com/fdaudens/status/1906345214089785634
“DeepSeek V3 0324 is also on WebDev Arena, and has ranked as #4! 🐳 👀 📈 HUGE improvement over past DeepSeek V3 (~250 score lift from 962 -> 1208). DeepSeek V3 0324 is the #1 open model! 🏆 https://x.com/lmarena_ai/status/1906784744358178932
“OpenArm is a fully open-source robotic arm designed for human manipulation data collection – developed by the Reazon Human Interaction Lab in Japan. V0.2 beta features smoother teleop with gravity compensation and bilateral force feedback. https://x.com/TheHumanoidHub/status/1907310437907177834
“xAI announced that its Grok chatbot is now integrated directly into messaging app Telegram Grok can be accessed as a bot with Telegram username “@GrokAI” However, it’s only available to Telegram premium users as of now! https://x.com/rowancheung/status/1905130285231927418
