https://tpsreport.news/news/openai-719-math-manuscripts-agmai-guidelines-lean-discrepancies TPS — Tokens Per Second en 2026-10-08T18:20:47.977Z Only 10 of OpenAI's 719 math manuscripts include chain of thought, falling short of expert guidelines OpenAI, mathematics, Lean, formal verification, AGMAI, Navier-Stokes, Terence Tao, AI research https://tpsreport.news/news/google-ai-edge-foresight-offline-meeting-transcription-mac TPS — Tokens Per Second en 2026-10-08T15:35:48.102Z Google releases free AI Edge Foresight, a Mac note-taking app that transcribes meetings fully offline Google, AI Edge Foresight, EmbeddingGemma, on-device AI, note-taking, macOS, Apple Silicon, privacy https://tpsreport.news/news/stepfun-step-5-preview-600b-moe-1m-context TPS — Tokens Per Second en 2026-10-08T12:50:49.812Z StepFun releases Step 5 Preview: 600B MoE with 1M context at $1/$2.70 per 1M tokens StepFun, Step 5 Preview, Mixture-of-Experts, agentic AI, long context, OpenRouter, model release https://tpsreport.news/news/google-cloud-gemini-agent-universal-agent-for-work TPS — Tokens Per Second en 2026-10-08T12:05:42.840Z Google Cloud launches Gemini agent, a multi-model enterprise agent that runs Gemini and Claude, in private preview Gemini, Google Cloud, Google Workspace, AI agents, enterprise AI, Claude, Gemini Enterprise https://tpsreport.news/news/gpt-6-chatgpt-intelligent-ui-sol-luna TPS — Tokens Per Second en 2026-10-07T19:20:50.898Z OpenAI launches GPT-6 in ChatGPT with 'Intelligent UI' and interactive answers; Sol for paid users, Luna for free GPT-6, OpenAI, ChatGPT, Intelligent UI, GPT-6 Sol, GPT-6 Luna, model release, Gemini https://tpsreport.news/news/claude-haiku-5-5-amazon-bedrock-effort-controls TPS — Tokens Per Second en 2026-10-07T19:05:59.581Z Claude Haiku 5.5 arrives on Amazon Bedrock; Anthropic claims ~75% lower cost than Haiku 4.5 Claude Haiku 5.5, Anthropic, Amazon Bedrock, AWS, Claude 5.5, subagents, effort controls, model availability https://tpsreport.news/news/chatgpt-for-teens-common-sense-media-unacceptable-risk TPS — Tokens Per Second en 2026-10-07T18:35:53.123Z Common Sense Media rates ChatGPT for Teens an 'unacceptable risk,' citing failures on 3 of 5 Red Lines OpenAI, ChatGPT for Teens, AI safety, Common Sense Media, teen safety, regulation, CHATBOT Act https://tpsreport.news/news/claude-haiku-5-5-75-percent-cheaper-than-haiku-4-5 TPS — Tokens Per Second en 2026-10-07T18:21:11.179Z Anthropic releases Claude Haiku 5.5, claims ~75% lower running cost than Haiku 4.5 Anthropic, Claude, Claude Haiku 5.5, small models, pricing, agents, GPT-5.5 Luna https://tpsreport.news/news/microsoft-copilot-windows-local-files-hybrid-intelligence TPS — Tokens Per Second en 2026-10-07T18:20:47.113Z Microsoft's Copilot gets access to local Windows files and OS-level actions under 'Hybrid Intelligence' Microsoft, Copilot, Windows, Hybrid Intelligence, AI agents, Autopilot, on-device AI https://tpsreport.news/news/microsoft-windows-search-copilot-chatbot-fall-2026 TPS — Tokens Per Second en 2026-10-07T18:06:30.220Z Microsoft will turn Windows Search into a Copilot-connected command interface this fall microsoft, windows, copilot, windows-search, surface, ai-agents, product-update https://tpsreport.news/news/anthropic-python-sdk-1-12-0-claude-haiku-5-5 TPS — Tokens Per Second en 2026-10-07T18:06:08.499Z Anthropic Python SDK 1.12.0 adds claude-haiku-5-5 and typed computer and browser tool calls Anthropic, Claude, Python SDK, claude-haiku-5-5, changelog, Managed Agents, computer use, API https://tpsreport.news/news/liquid-ai-d1-3b-open-decision-model-edge TPS — Tokens Per Second en 2026-10-07T17:06:01.536Z Liquid AI releases open d1-3B decision model: 16 ms on Jetson AGX Thor, 48.57 on Decision Index Liquid AI, d1-3B, d1-omni-600M, decision models, edge AI, open weights, multimodal, NVIDIA Jetson https://tpsreport.news/news/openai-decisions-api-gpt-6-luna-public-beta TPS — Tokens Per Second en 2026-10-07T15:50:56.916Z OpenAI's Decisions API enters public beta at $0.10 per 1M input tokens, with free output OpenAI, Decisions API, gpt-6-luna, API pricing, classification, API tiers, HIPAA https://tpsreport.news/news/google-synthid-detector-global-launch-apple-openai-nvidia-kakao TPS — Tokens Per Second en 2026-10-07T15:21:09.927Z Google's SynthID detector goes global in English; Apple Intelligence watermarking to follow 'soon' SynthID, Google, AI watermarking, AI detection, Apple Intelligence, OpenAI, Nvidia, Kakao https://tpsreport.news/news/grok-bot-claude-opus-5-5-backend-upgrade TPS — Tokens Per Second en 2026-10-07T15:20:49.605Z Grok Bot now runs on Anthropic's Claude Opus 5.5, replacing SpaceXAI/Cursor models for core intelligence Grok Bot, SpaceXAI, Anthropic, Claude Opus 5.5, AI agents, Meta Muse, ChatGPT Dot, xAI https://tpsreport.news/news/google-playground-text-to-game-gemini-nano-banana-lyria TPS — Tokens Per Second en 2026-10-07T14:50:56.788Z Google launches Playground, a free text-to-game platform for US adults built on Gemini, Nano Banana and Lyria Google, Gemini, Nano Banana, Lyria, Playground, Unity Spark, game development, generative AI https://tpsreport.news/news/meta-muse-ipad-app-launch-connectors TPS — Tokens Per Second en 2026-10-07T14:35:45.069Z Meta launches Muse AI agent app for iPad after weeks as the top free iPhone app in the US Meta, Muse, iPad, AI agents, App Store, connectors, agentic AI, product update https://tpsreport.news/news/falcon-asr-1-6b-arabic-emirati-speech-recognition TPS — Tokens Per Second en 2026-10-07T13:35:55.630Z TII releases 1.6B Falcon-ASR, claims 20.92% Arabic WER against best listed 23.17% Falcon-ASR, TII, speech-recognition, Arabic, Emirati dialect, ASR, Hugging Face, multilingual https://tpsreport.news/news/unsloth-embeddinggemma-2-gguf-740m-multimodal-embedding TPS — Tokens Per Second en 2026-10-07T13:21:03.653Z Unsloth publishes GGUF quantizations of Google DeepMind's 740M-parameter EmbeddingGemma 2 multimodal embedding model EmbeddingGemma 2, Unsloth, Google DeepMind, GGUF, embeddings, multimodal, quantization, open-weight https://tpsreport.news/news/nvidia-nemotron-3-ultra-ioi-imo-2026-gold TPS — Tokens Per Second en 2026-10-07T13:05:50.438Z NVIDIA's fine-tuned Nemotron 3 Ultra scores 535.4/600 at IOI 2026 and 30/42 at IMO 2026, both above gold nvidia, nemotron, IOI, IMO, reasoning, fine-tuning, competitive-programming, open-weights https://tpsreport.news/news/anthropic-cyber-verification-program-expansion-claude TPS — Tokens Per Second en 2026-10-07T12:20:44.229Z Anthropic Opens Cyber Verification Program to More Security Teams With Reduced Claude Safety Filters Anthropic, Claude, cybersecurity, Cyber Verification Program, Project Glasswing, AI safety, penetration testing https://tpsreport.news/news/openai-372-ai-generated-math-results-github TPS — Tokens Per Second en 2026-10-07T09:05:48.479Z OpenAI publishes 372 AI-generated math results on GitHub, claims they solve or advance open problems OpenAI, mathematics, Lean, formal verification, Riemann hypothesis, Navier-Stokes, AI research, ChatGPT Pro https://tpsreport.news/news/cline-v4-1-23-claude-base-url-fix-provider-defaults TPS — Tokens Per Second en 2026-10-07T07:35:49.299Z Cline v4.1.23 fixes Claude 400 errors on custom Anthropic URLs, changes defaults for 19 providers Cline, release-notes, coding-agents, MCP, Claude, Azure AI Foundry, GPT-6.1 Sol, Kimi K3 https://tpsreport.news/news/cline-cli-v3-0-69-mcp-startup-timeout-10-seconds TPS — Tokens Per Second en 2026-10-07T06:20:43.236Z Cline CLI v3.0.69 raises MCP startup timeout from 3 to 10 seconds, fixing silently dropped Windows servers Cline, CLI, MCP, Windows, Anthropic, Kimi K3, Azure AI Foundry, coding agents https://tpsreport.news/news/llm-mistral-0-16-reasoning-model-support-mistral-large-4 TPS — Tokens Per Second en 2026-10-06T21:50:43.308Z llm-mistral 0.16 adds reasoning model support, including newly released Mistral Large 4 llm-mistral, Mistral Large 4, LLM CLI, Simon Willison, reasoning models, Mistral API, plugins https://tpsreport.news/news/musubi-policylm-1-7b-open-weights-moderation-decision-model TPS — Tokens Per Second en 2026-10-06T20:50:49.142Z Musubi Releases Open-Weights PolicyLM-1.7B, a Sub-50ms Content Moderation Model PolicyLM-1.7B, Musubi, content moderation, decision models, open weights, small language models, trust and safety https://tpsreport.news/news/claude-google-workspace-add-on-docs-sheets-slides-sidebar TPS — Tokens Per Second en 2026-10-06T19:20:45.375Z Anthropic launches Claude for Google Workspace add-on in public beta, adding sidebars to Docs, Sheets and Slides Anthropic, Claude, Google Workspace, Google Docs, Google Sheets, Google Slides, product update, integrations