Epoch's IPI Shows Nvidia GB300 Delivers 25x an H100's Real Inference Capacity
Epoch AI's new Inference Performance Index scores chips by memory and interconnect, not just FLOP/s, revealing GB300 serves 25x H100's load.
20 stories
Epoch AI's new Inference Performance Index scores chips by memory and interconnect, not just FLOP/s, revealing GB300 serves 25x H100's load.
Anthropic's new frontier models formalized Fermat's Last Theorem in 30 hours, scored 49.1% on FrontierCode 2.0, and dropped prices 20%.
Meta's new efficient-tier model matches Claude Haiku 5.5 on intelligence at a third of the cost, with 1M context and native multimodal input.
World Labs' Atlas took the top spot on Rapidata's camera movement leaderboard, beating Google, Alibaba, and NVIDIA models across 513K human votes.
Meta's new Muse Ember matches Claude Haiku 5.5 at a third the cost, with 1M context and native multimodal support for agent work.
Nous Research puts the full Hermes Desktop interface in a browser tab, letting cloud-hosted agents run 24/7 and stay reachable from any device.
ByteDance's new image model brings native transparent output, 100-round stable editing, and sharper text rendering, now available through Higgsfield.
A new splatting method extends 3D Gaussian Splatting from points to connected curves, producing strands that work for hair, wires, and physics simulation.
Google wired its Maps Platform into AI Studio, letting you prompt Gemini to spin up full geospatial apps with no billing setup or key juggling.
Latent Sparse Attention (LSA) from Prefrontal AI and ETH Zurich clusters keys to speed long-context LLM inference, cutting 10M-token time to first token by over 9 times
Meta's LLM-powered test generator writes failing tests that prove bugs exist, catching 74 privacy incidents and cutting privacy SEVs 20 percent.
UkisAI released mixed-precision GGUF quantizations of Swift Flash Next that cut thinking tokens by 63% while keeping accuracy loss under 1%.
Brett Adcock's AI assistant now lives in your text messages, bringing agentic browsing, logins, and payments to iMessage and RCS without opening an app.
NVIDIA's 16B omnimodal world model unifies text, image, video, audio, and robot action generation under one Mixture-of-Transformers architecture.
A free Windows app pipes video into editable 3D character animation locally, bundling six human-pose AI models with FBX, BVH, VMD and Blender export.
Baidu's new automated research system modifies code, runs experiments, and keeps only verified wins, reaching SOTA on NanoGPT Speedrun and NanoChat AutoResearch.
A new ACL 2026 method called CLaRE predicts which facts will break when you edit an LLM, using forward activations instead of gradients.
A new method discovers compact, interpretable polynomial learning rules per layer, offering a symbolic alternative to backpropagation for small networks.
Pinky is an open-source MCP gateway that lets Claude Code, Codex and Cursor drive a real iPhone or Android phone with human approval gates baked in.
Sarvam AI brings on former Google Cloud India MD Sashikumar Sreedharan as President and CBO to scale its sovereign AI push with enterprises and government.