https://tpsreport.news/news/openai-biobasicsbench-wet-lab-agent-benchmark TPS: Tokens Per Second en 2026-11-06T19:20:39.809Z OpenAI introduces BioBasicsBench, a 100-task benchmark for AI agents in everyday wet-lab science OpenAI, BioBasicsBench, benchmark, AI agents, wet-lab, biology https://tpsreport.news/news/anthropic-model-4-aeci-178-metr-report TPS: Tokens Per Second en 2026-11-06T18:35:55.392Z METR report: Anthropic's unreleased Model 4 scores 178.6 on AECI, ahead of Opus 5.5 Anthropic, METR, AECI, Opus 5.5, Model 4, benchmark, AI R&D, Responsible Scaling Policy https://tpsreport.news/news/meta-muse-spaces-custom-app-experiences TPS: Tokens Per Second en 2026-11-06T18:20:51.507Z Meta adds "spaces" to Muse AI, letting users build app-like hubs by describing them Meta, Muse, Muse Wonder, AI assistant, spaces, product update https://tpsreport.news/news/salesforce-koa-decisions-gemma-image-audio TPS: Tokens Per Second en 2026-11-06T16:50:48.763Z Salesforce launches Koa Decisions, a Gemma-based model that classifies text, image and audio requests Salesforce, Koa Decisions, HyperClassifier, Gemma, Agentforce, multimodal https://tpsreport.news/news/google-antigravity-2-29-remote-agents-voice-mode TPS: Tokens Per Second en 2026-11-06T15:50:49.819Z Google Antigravity 2.29 adds cloud Remote Agents for Google AI Pro and Ultra subscribers Google, Antigravity, Remote Agents, Voice Mode, Gemini Live, agentic coding, Google AI Pro, Google AI Ultra https://tpsreport.news/news/underdog-woof-31m-parameter-open-text-to-speech-model TPS: Tokens Per Second en 2026-11-06T13:21:02.001Z Underdog releases Woof, a 31M-parameter open text-to-speech model that runs in a browser Underdog, Woof, text-to-speech, on-device AI, open models, edge AI, local AI, Bark https://tpsreport.news/news/perplexity-teammates-ai-agents-max-200-month TPS: Tokens Per Second en 2026-11-06T11:50:45.638Z Perplexity launches 'teammates' AI agents inside Computer, exclusive to $200-a-month Max plan Perplexity, AI agents, Perplexity Computer, Perplexity Max, teammates, agentic AI, product update https://tpsreport.news/news/cola2-bytedance-seed-continuous-latent-9b TPS: Tokens Per Second en 2026-11-06T10:20:56.629Z ByteDance Seed releases Cola2, a 9B continuous-latent language model with 60.10 eight-task average Cola2, ByteDance Seed, ByteDance, continuous latent, diffusion transformer, open weights, base model, Hugging Face https://tpsreport.news/news/open-asr-leaderboard-egyptian-nigerian-english-liva-ai TPS: Tokens Per Second en 2026-11-06T10:05:46.318Z Open ASR Leaderboard adds private Egyptian and Nigerian English test sets from Liva AI Open ASR Leaderboard, Hugging Face, Liva AI, ASR, speech recognition, benchmark, accents https://tpsreport.news/news/sage-sci-perception-scientific-encoders-language-model TPS: Tokens Per Second en 2026-11-06T09:20:46.522Z Sage Sci Perception links scientific encoders to a language model; ECG, sleep and RNA results published Sage Sci Perception, internlm, scientific AI, multimodal, Hugging Face, protein, spectra, benchmarks https://tpsreport.news/news/mnn-pocket-0-3b-173mb-on-device-model TPS: Tokens Per Second en 2026-11-06T08:08:37.356Z Taobao's MNN team open-sources MNN-Pocket-0.3B, a 300M-parameter on-device model with a 173MB INT4 package MNN-Pocket-0.3B, Taobao, Alibaba, on-device AI, small language model, open source, MNN, Hugging Face https://tpsreport.news/news/ant-ling-3-2-flash-560b-multimodal-1m-context TPS: Tokens Per Second en 2026-11-06T08:08:00.047Z Ant Group releases Ling-3.2-flash: 560B-parameter multimodal model with 1M-token context Ant Group, Ling-3.2-flash, multimodal, long context, agents, model release https://tpsreport.news/news/360-tulongfeng-cybergym-99-4-percent TPS: Tokens Per Second en 2026-11-06T08:07:29.274Z 360's Tulongfeng tops CyberGym with 99.4% vulnerability reproduction rate across 1,507 tasks CyberGym, Tulongfeng, 360, AI security, vulnerability discovery, benchmark, agents https://tpsreport.news/news/jimeng-short-drama-studio-seedance-2-5-pricing TPS: Tokens Per Second en 2026-11-06T08:06:30.380Z Jimeng Short Drama Studio links script, assets and storyboards; Seedance 2.5 as low as 0.15 yuan per second Jimeng, Seedance 2.5, AI video, short drama, DreamScript, Jianying, Qbitai https://tpsreport.news/news/tencent-hy4-tenfold-capacity-workbuddy-demand TPS: Tokens Per Second en 2026-11-06T07:20:48.499Z Tencent expands Hy4 inference cluster tenfold after WorkBuddy usage surge Tencent, Hy4, WorkBuddy, Artificial Analysis, Agent Arena, inference capacity, benchmarks https://tpsreport.news/news/robber-bots-ai-vending-machine-profit-maximization-study TPS: Tokens Per Second en 2026-11-06T03:50:51.451Z Harvard-led study: AI agents told only to maximize profit lied on refunds and colluded in vending simulation AI agents, Andon Labs, Harvard Business School, AI safety, benchmark, simulation, Claude Opus 4.6, Gemini 2.5 Pro https://tpsreport.news/news/deepseek-harness-1-54-million-daily-users-plugin-marketplace TPS: Tokens Per Second en 2026-11-06T02:05:53.421Z DeepSeek Harness Reaches 1.54M Daily Users, Plans Official Plugin Marketplace DeepSeek, DeepSeek Harness, agent runtime, plugins, open source, plugin marketplace https://tpsreport.news/news/inkling-2-small-thinking-machines-lab-594b-moe TPS: Tokens Per Second en 2026-11-06T01:06:20.751Z Thinking Machines Lab launches Inkling 2 Small, a 594B-parameter sparse MoE model, on Tinker Inkling 2 Small, Thinking Machines Lab, Tinker, Mixture-of-Experts, closed weights, model release https://tpsreport.news/news/inkling-2-small-thinking-machines-intelligence-index-31 TPS: Tokens Per Second en 2026-11-06T01:05:55.451Z Thinking Machines' Inkling 2 Small scores 31 on Artificial Analysis Intelligence Index, matching Gemini 3.8 Flash (high) Inkling 2 Small, Thinking Machines, Artificial Analysis, Intelligence Index, open weights, benchmark, Tinker https://tpsreport.news/news/perplexity-agent-api-lower-costs-ultra-effort TPS: Tokens Per Second en 2026-11-06T00:20:54.225Z Perplexity cuts per-query cost of every Agent API effort level and adds an ultra tier Perplexity, Agent API, BrowseComp, DSQA, pricing, research agents https://tpsreport.news/news/cursor-live-apps-cloud-agents TPS: Tokens Per Second en 2026-11-05T22:36:46.929Z Cursor adds live app building: cloud agents now deploy apps with database, auth and hosting Cursor, Anysphere, cloud agents, app deployment, AI coding, product update https://tpsreport.news/news/metr-anthropic-redacted-evidence-100-page-report TPS: Tokens Per Second en 2026-11-05T21:50:42.644Z METR says Anthropic redacted most of the evidence it found most informative in its ~100-page report METR, Anthropic, AI evaluation, transparency, redaction, AI safety https://tpsreport.news/news/world-labs-frameboy-real-time-atlas-15-fps TPS: Tokens Per Second en 2026-11-05T19:51:25.991Z World Labs Releases Frameboy, Streaming Atlas Scenes to Browsers at Up to 15 FPS World Labs, Frameboy, Atlas, Viewfinder, 3D scenes, real-time inference, research preview https://tpsreport.news/news/nvidia-cosmos-3-nano-open-world-simulation-checkpoints TPS: Tokens Per Second en 2026-11-05T19:51:04.799Z NVIDIA releases four open-weight Cosmos 3 Nano checkpoints, 16B parameters each, for driving and robot simulation NVIDIA, Cosmos 3, world models, autonomous driving, robotics, open-weight, simulation, LiDAR https://tpsreport.news/news/ai-costs-falling-data-center-backlash-gpt-6-1-sol TPS: Tokens Per Second en 2026-11-05T19:36:09.127Z AI Running Costs Fell This Year Even as Server Prices Rose, but Data Center Curbs Test the Trend data centers, AI pricing, GPT-6.1 Sol, GPT-6 Astra, Anthropic, Amazon Web Services, Artificial Analysis, compute costs https://tpsreport.news/news/meta-muse-voice-mode-expressive-avatars TPS: Tokens Per Second en 2026-11-05T19:21:26.117Z Meta adds real-time voice mode and expressive avatars to Muse AI agent in US app Meta, Muse, voice mode, AI agent, avatars, AI glasses, Muse Charm https://tpsreport.news/news/gemini-canvas-sketch-to-app-prototyping-demo TPS: Tokens Per Second en 2026-11-05T18:21:27.395Z Google shows Gemini Canvas turning a photo of a paper sketch into an interactive app Google, Gemini, Gemini Canvas, app prototyping, multimodal, vibe coding https://tpsreport.news/news/metr-anthropic-ai-development-pace-doubled TPS: Tokens Per Second en 2026-11-05T18:21:08.485Z METR Report: Anthropic's AI Development Pace May Have Doubled Due to Internal AI Use Anthropic, METR, AI R&D, AI progress, analysis https://tpsreport.news/news/claude-personal-finance-plaid-read-only TPS: Tokens Per Second en 2026-11-05T18:20:51.755Z Anthropic launches Personal Finance in Claude with read-only Plaid access for US Pro, Max and Ultra users Anthropic, Claude, Personal Finance, Plaid, privacy, ChatGPT Finances, product update https://tpsreport.news/news/gemini-omni-1-2-flash-video-generation TPS: Tokens Per Second en 2026-11-05T17:36:36.302Z Google releases Gemini Omni 1.2 Flash video model: 15-second clips up to 4K with native audio Gemini Omni 1.2 Flash, Google, video generation, multimodal, OpenRouter, Google AI Studio, 4K video https://tpsreport.news/news/gemini-omni-1-1-flash-video-generation-4k TPS: Tokens Per Second en 2026-11-05T17:36:08.230Z Google releases Gemini Omni 1.1 Flash, a generally available video model with 4K output and native audio Gemini Omni 1.1 Flash, Google, video generation, OpenRouter, multimodal, native audio, 4K video https://tpsreport.news/news/openai-gpt-6-cyber-critical-exploitgym TPS: Tokens Per Second en 2026-11-05T17:20:54.036Z OpenAI launches GPT-6 Cyber, rated Critical, with 48.5% on ExploitGym OpenAI, GPT-6 Cyber, cybersecurity, ExploitGym, Preparedness Framework, Daybreak Red, AI safety https://tpsreport.news/news/apple-siri-ai-waitlist-removed-ios-27-beta TPS: Tokens Per Second en 2026-11-05T16:50:48.544Z Apple removes Siri AI waitlist in iOS 27 beta; compatible devices now download it immediately Apple, Siri AI, iOS 27, Apple Intelligence, beta, virtual assistant https://tpsreport.news/news/openai-gpt-6-astra-safety-research-underperformance TPS: Tokens Per Second en 2026-11-05T14:20:57.196Z OpenAI: GPT-6 Astra completed about 40% fewer safety research tasks than identical general tasks OpenAI, GPT-6 Astra, AI safety, alignment, sandbagging, reinforcement learning, GPT-5.6 Sol https://tpsreport.news/news/openrouter-auto-code-coding-agent-router TPS: Tokens Per Second en 2026-11-05T13:20:56.601Z OpenRouter launches Auto Code, a router that picks models for coding agents by pass rate, time and cost per solved task OpenRouter, Auto Code, model routing, coding agents, AI routers, developer tools https://tpsreport.news/news/visa-ai-shopping-agents-study-57-percent-success TPS: Tokens Per Second en 2026-11-05T13:05:47.036Z Visa test: 22 AI shopping agents succeeded about 57% of the time; Gemini and Meta agents top at 82% AI agents, agentic commerce, Visa, benchmark, Gemini, Claude, OpenAI, Meta https://tpsreport.news/news/gradium-phonon-on-device-tts-200000-downloads TPS: Tokens Per Second en 2026-11-05T11:05:46.119Z Gradium says Phonon on-device TTS passed 200,000 downloads in two weeks Gradium, Phonon, text-to-speech, on-device, Open TTS Leaderboard, Hugging Face https://tpsreport.news/news/openai-gpt-image-2-6-flare-sunburst TPS: Tokens Per Second en 2026-11-05T08:21:42.354Z OpenAI releases GPT-Image-2.6 Flare and Sunburst, adding GIF output and SVG export OpenAI, GPT-Image-2.6, Flare, Sunburst, image generation, SVG, GIF, API https://tpsreport.news/news/baidu-ernie-5-5-flash-base-open-source-1m-context TPS: Tokens Per Second en 2026-11-05T08:21:21.657Z Baidu open-sources ERNIE 5.5 Flash Base, an MoE model with 1M-token context for fine-tuning and inference ERNIE 5.5, Baidu, open-source, MoE, base model, long context, Hugging Face https://tpsreport.news/news/superclue-office-october-2026-doubao-work-kimi-work-workbuddy TPS: Tokens Per Second en 2026-11-05T07:35:51.481Z Doubao Work tops SuperCLUE-Office October benchmark at 85.33, ahead of Kimi Work (84.57) SuperCLUE, office agents, Doubao Work, Kimi Work, WorkBuddy, benchmark, ByteDance, Moonshot AI https://tpsreport.news/news/moonshot-ai-kimi-hosted-agents-harness-as-a-service TPS: Tokens Per Second en 2026-11-05T07:20:47.626Z Moonshot AI launches Kimi Hosted Agents, a managed agent runtime with sandboxes at 0.072 yuan per hour Moonshot AI, Kimi, Kimi Hosted Agents, AI agents, agent harness, MCP, enterprise AI, Kimi K3.1 Pro https://tpsreport.news/news/openai-python-3-46-0-gpt-6-rosalind-residency TPS: Tokens Per Second en 2026-11-05T06:35:42.876Z OpenAI Python SDK 3.46.0 adds GPT-6 Rosalind to model catalog and Australia, India, Japan residency options OpenAI, openai-python, SDK, GPT-6 Rosalind, data residency, changelog https://tpsreport.news/news/meta-stops-fruit-codenames-watermelon-wonder-full TPS: Tokens Per Second en 2026-11-05T04:50:43.417Z Meta stops using fruit codenames for AI models after "Watermelon," the internal name for Wonder Full Meta, Wonder Full, Watermelon, codenames, Mark Zuckerberg, Alexandr Wang, frontier model https://tpsreport.news/news/plamo-3-translate-open-weights-0-8b-31b TPS: Tokens Per Second en 2026-11-05T04:20:46.637Z Preferred Networks opens PLaMo 3 Translate weights in four sizes, 0.8B to 31B PLaMo 3 Translate, Preferred Networks, open weights, translation, on-device, Hugging Face https://tpsreport.news/news/moxin-moworld-2-0-ascend-native-4d-world-model TPS: Tokens Per Second en 2026-11-05T02:51:19.601Z Moxin Technology launches MoWorld 2.0, a 4D world model trained and deployed natively on Huawei Ascend MoWorld 2.0, Moxin Technology, world model, Huawei Ascend, 4D Gaussian, physics simulation, embodied AI https://tpsreport.news/news/huawei-xiaoyi-work-openpangu-2-1-pro-harmonyos TPS: Tokens Per Second en 2026-11-05T02:50:55.642Z Huawei's Xiaoyi Work is first to run openPangu-2.1-Pro, labeled for HarmonyOS task enhancement Huawei, openPangu, Pangu, Xiaoyi Work, HarmonyOS, AI agents, Huawei Cloud, Ascend https://tpsreport.news/news/webai-twil-8b-reasoning-model-math-500 TPS: Tokens Per Second en 2026-11-04T23:05:58.606Z webAI Releases TwIL-8B, an 8.8B Reasoning Model With 98.2 on MATH-500 webAI, TwIL-8B, reasoning model, IBM Granite, MATH-500, LiveCodeBench, 8B https://tpsreport.news/news/ricursive-ripple-65b-preview-terminal-bench-80-5 TPS: Tokens Per Second en 2026-11-04T23:05:43.639Z Ricursive releases Ripple-65B Preview, 65B dense open-weight model, scoring 80.5 on Terminal-Bench 2.1 Ripple-65B, Ricursive, open-weight, Apache 2.0, Terminal-Bench 2.1, dense model https://tpsreport.news/news/openai-autonomous-pipeline-948-linux-kernel-privilege-escalation-vulnerabilities TPS: Tokens Per Second en 2026-11-04T22:36:06.223Z OpenAI pipeline finds, validates and patches 948 Linux kernel privilege-escalation flaws OpenAI, Linux kernel, security, vulnerabilities, privilege escalation, autonomous agents https://tpsreport.news/news/perplexity-decisions-api-live-web-grounding TPS: Tokens Per Second en 2026-11-04T21:35:52.329Z Perplexity adds optional live web grounding to its Decisions API, available immediately Perplexity, Decisions API, web grounding, API, changelog