Image created with gemini-2.5-flash-image with claude-sonnet-4-5. Image prompt: An 80s suburban street at autumn dusk, warm streetlight glow, a child in a ghost costume holding up a hand-drawn Halloween trick-or-treat route map with houses marked and circled, fallen leaves on sidewalk, jack-o-lanterns glowing on porches in background, nostalgic film grain
Meta’s $75B AI Infrastructure Bet: Inside the Biggest Cloud Deals of 2025 https://allenarch.dev/blog/meta-75b-ai-infrastructure-bet/
The Building Blocks of Agentic AI: From Kernels to Clusters https://ai.meta.com/blog/introducing-pytorch-native-agentic-stack/
You can capture reality with remarkable fidelity – here’s a 1:1 3d gaussian splat perfectly aligned with reality. Once Meta layers in a scalable version of their codec avatar tech – this stuff is about to get trippy. Literally transcending time & space. https://x.com/bilawalsidhu/status/1982441732186022219
ollama run qwen3-vl Ollama’s engine now supports all the Qwen 3 VL models locally. 2B to 235B parameter sizes. The smaller models work exceptionally well for their size. The latest version of Ollama v0.12.7 is needed! Give it a try! 👇👇👇 https://x.com/ollama/status/1983683646864126155
Meta presents SPICE Self-Play In Corpus Environments Improves Reasoning https://x.com/_akhaliq/status/1983565708098146809
Brain-IT: Image Reconstruction from fMRI via Brain-Interaction Transformer A new paper from Weizmann Institute of Science, getting reconstructions that are not complete nonsense from ONLY 15 min of data. We previously demonstrated SOTA for 1 hr of data with MindEye2. This https://x.com/iScienceLuvr/status/1984195725253804449
Clearly NVIDIA’s nccl isn’t good enough for scaling up as there are new alternative collective comms libraries being developed Meta has published a paper on NCCLX Collective Communication for 100k+ GPUs https://x.com/StasBekman/status/1982861472024932409
PewDiePie in 2025: – built a 10×4090 rig – runs Llama 70B, gpt-oss-120B & Qwen 245B locally via vLLM – built a custom web UI (chat, RAG, search, TTS) – ran protein-folding simulations for charity – created an AI “council”, a swarm of 64 models – now fine-tuning his own model https://x.com/Yuchenj_UW/status/1984309989134254493
Yann LeCun says none of the humanoid robot companies has any idea how to make the robots smart enough to be generally useful, and their future hinges on breakthroughs in “”world model planning type architectures.”” https://x.com/TheHumanoidHub/status/1981949426198360137
