https://tpsreport.news/news/gemini-app-gems-skills-migration-november-2026 TPS — Tokens Per Second en 2026-09-27T17:35:56.474Z Google to Replace Gems with Skills in Gemini App Starting November 17, 2026 Gemini, Google, Gems, Skills, Gemini Spark, product update https://tpsreport.news/news/meta-muse-ai-agent-subscription-economy TPS — Tokens Per Second en 2026-09-27T14:50:57.742Z Meta's Muse AI Agent Targets $1,887 Average Annual Subscription Spend, Threatens Recurring-Revenue Business Model Meta, Muse, AI agent, subscription economy, personal assistant, fintech, consumer AI https://tpsreport.news/news/nvidia-nemotron-3-diarization-speaker-id-model TPS — Tokens Per Second en 2026-09-27T11:20:49.051Z Nvidia Releases Nemotron 3 Diarization, a Free 100M-Parameter Model That Tracks 8 Speakers in Real Time Nvidia, speaker diarization, open source AI, speech recognition, Nemotron, audio AI https://tpsreport.news/news/gpt-6-astra-homebody-robot-kitchen-cleanup TPS — Tokens Per Second en 2026-09-27T11:05:49.455Z Stanford, Caltech Researchers Wire GPT-6 Astra Directly Into a Robot to Clean an Unfamiliar Kitchen GPT-6 Astra, robotics, OpenAI, Stanford, Caltech, Unitree G1, vision-language models, Isaac Sim https://tpsreport.news/news/apple-lensvlm-9b-vision-language-model TPS — Tokens Per Second en 2026-09-26T19:20:55.635Z Apple Releases LensVLM-9B, a 9B Vision-Language Model That Selectively Decompresses Text Images Apple, Vision Language Model, VLM, Qwen, open weights, document AI, research https://tpsreport.news/news/ai-access-eliminates-i-dont-know-study TPS — Tokens Per Second en 2026-09-26T17:05:48.982Z Study: Access to AI Advice Nearly Eliminates People's Willingness to Say "I Don't Know" AI research, human-AI interaction, cognitive science, study, AI trust, epistemics https://tpsreport.news/news/openai-pauses-training-most-capable-models TPS — Tokens Per Second en 2026-09-26T16:50:50.847Z OpenAI Pauses Training of Its Most Capable Models After AI Escapes Sandbox OpenAI, AI safety, AI security, model training, sandbox escape, AI agents https://tpsreport.news/news/nvidia-sol-pi-coding-agent-harness-optimization TPS — Tokens Per Second en 2026-09-26T10:50:55.213Z Nvidia's SoL-Pi Cuts Coding Agent Token Usage by Up to 49% Through Automated Harness Optimization nvidia, coding-agents, ai-agents, token-efficiency, harness-optimization, codex, claude-code, research https://tpsreport.news/news/gpt-6-astra-furniture-assembly-benchmark-80-percent TPS — Tokens Per Second en 2026-09-26T09:50:48.075Z OpenAI's GPT-6 Astra Scores 80% on IKEA Assembly-Error Benchmark, Up From 28% Ten Months Ago OpenAI, GPT-6 Astra, benchmark, multimodal, Epoch AI, Claude, visual reasoning https://tpsreport.news/news/meta-muse-glimmer-30b-open-weight-agentic-model TPS — Tokens Per Second en 2026-09-26T06:20:53.640Z Meta Releases Muse Glimmer 30B, an Open-Weight Agentic Model for Consumer Hardware Meta, Muse Glimmer, open-weight, agentic AI, multimodal, OpenRouter https://tpsreport.news/news/ai-sdk-openai-4-0-78-gpt-6-reasoning-effort TPS — Tokens Per Second en 2026-09-26T02:05:51.598Z Vercel AI SDK Patch Adds Support for 'None' Reasoning Effort on GPT-6 Sol and Luna ai-sdk, vercel, openai, gpt-6, changelog, reasoning-effort, sdk-update https://tpsreport.news/news/meta-muse-early-access-program-features TPS — Tokens Per Second en 2026-09-25T20:50:46.831Z Meta Opens Early Access Signups for New Muse AI Features Via In-App Prompt Meta, Muse, AI assistant, early access, Connect 2026, AI glasses