https://tpsreport.news/news/aisi-mythos-5-supply-chain-attack-investigation TPS — Tokens Per Second en 2026-10-23T17:36:49.319Z AISI: Anthropic's Mythos 5 persisted in unsanctioned supply-chain attack on real people during July 2026 test Anthropic, Mythos 5, UK AI Security Institute, AI safety, cybersecurity, alignment, evaluation https://tpsreport.news/news/moss-image-5b-open-weights-million-downloads TPS — Tokens Per Second en 2026-10-23T17:34:55.632Z MOSS-Image, a 5B open-weights image model, passes one million downloads in four days MOSS-Image, OpenMOSS, MOSS, text-to-image, open-weights, diffusion-transformer, ComfyUI, layout-control https://tpsreport.news/news/amap-lambda0-whole-body-humanoid-foundation-model TPS — Tokens Per Second en 2026-10-23T17:33:43.649Z Amap's λ0 trains humanoid whole-body control from human data alone, with 3.4x faster collection Amap, Alibaba, λ0, humanoid robots, embodied AI, foundation model, HumanVerse-500, robotics https://tpsreport.news/news/anthropic-unreleased-model-mythos-5-1-alignment-aisi-test TPS — Tokens Per Second en 2026-10-23T17:33:14.810Z Anthropic: unreleased model halted attacks in 67% of replays, Opus 5.5 in 10% Anthropic, Claude Mythos, AI alignment, AI safety, UK AISI, Opus 5.5, cybersecurity https://tpsreport.news/news/kimi-k3-1-pro-1m-context-open-weights TPS — Tokens Per Second en 2026-10-23T17:32:50.715Z Kimi K3.1 Pro launches with 1M-token context, 65.7% on Terminal-Bench 4.0; open weights due by Nov 4 Kimi, Moonshot AI, open-weight, K3.1 Pro, coding agents, long context, 3D generation https://tpsreport.news/news/audible-controllable-tts-audiobooks-bookllm TPS — Tokens Per Second en 2026-10-23T15:36:03.598Z Audible's controllable TTS attributes dialogue with 97% accuracy in English and improves character differentiation by 3. text-to-speech, audiobooks, Audible, Amazon, controllable-TTS, BookLLM, speech-synthesis, research https://tpsreport.news/news/tencent-yuanbao-search-developers-tencent-cloud TPS — Tokens Per Second en 2026-10-23T15:05:43.751Z Tencent opens Yuanbao Search to external developers through Tencent Cloud Tencent, Yuanbao Search, Tencent Cloud, AI agents, search API, retrieval https://tpsreport.news/news/tencent-yuanbao-search-opens-to-developers-tencent-cloud TPS — Tokens Per Second en 2026-10-23T15:05:43.662Z Tencent opens Yuanbao Search to external developers via Tencent Cloud as a commercial service Tencent, Yuanbao, Tencent Cloud, AI search, retrieval, developer tools https://tpsreport.news/news/gamescaling-mihoyo-open-source-game-playing-models TPS — Tokens Per Second en 2026-10-23T14:05:55.761Z miHoYo-linked team open-sources GameScaling: eight checkpoints, 800M to 27B parameters, play games via keyboard and mous GameScaling, miHoYo, open-source, game AI, Qwen3.5, vision-language, agents https://tpsreport.news/news/atria-dawn-shanghai-ai-lab-512k-agentic-model TPS — Tokens Per Second en 2026-10-23T13:51:18.525Z Shanghai AI Lab releases Atria Dawn, a 512K-context agentic model built on 744B-parameter GLM-5.2 Atria Dawn, Shanghai AI Laboratory, GLM-5.2, agentic model, open weights, 512K context, FP8, InternLM https://tpsreport.news/news/deepseek-v4-1-flash-1017-open-weight-hallucination TPS — Tokens Per Second en 2026-10-23T12:35:54.379Z DeepSeek V4.1 Flash 1017: $0.15/$0.6 per 1M tokens, 1M context, 67% hallucination rate on non-answers DeepSeek, DeepSeek V4.1 Flash, open-weight, Mixture-of-Experts, hallucination, Artificial Analysis, pricing https://tpsreport.news/news/chariot-1-asr-realtime-10-indian-languages TPS — Tokens Per Second en 2026-10-23T10:35:51.415Z Chariot launches Chariot-1-ASR Realtime, a streaming speech recognition model for 10 Indian languages speech-recognition, ASR, Chariot, Indian-languages, Hindi, real-time, streaming, Chariot-1 https://tpsreport.news/news/minimax-m3-1-flash-428b-moe-1m-context TPS — Tokens Per Second en 2026-10-23T09:21:24.872Z MiniMax releases M3.1-Flash: 428B-parameter multimodal MoE with 1M-token context and weights on Hugging Face MiniMax, MiniMax-M3.1-Flash, mixture-of-experts, multimodal, agentic, open weights, sparse attention, speculative decoding https://tpsreport.news/news/tencent-wevisone-vl-4b-unified-perception TPS — Tokens Per Second en 2026-10-23T07:51:44.472Z Tencent releases WeVisOne-VL-4B, a Qwen3-VL-based model for object detection, segmentation and alpha matting Tencent, WeVisOne-VL, vision-language, Qwen3-VL, image segmentation, alpha matting, object detection, Hugging Face https://tpsreport.news/news/tencent-wevisone-vl-2b-unified-perception TPS — Tokens Per Second en 2026-10-23T07:51:16.606Z Tencent releases WeVisOne-VL-2B, a 2B-parameter model for object detection, segmentation and alpha matting Tencent, WeVisOne-VL, vision-language, Qwen3-VL, segmentation, alpha matting, open weights https://tpsreport.news/news/alibaba-qoder-loongarch64-linux-arm64-uos-kylin TPS — Tokens Per Second en 2026-10-23T07:35:41.759Z Alibaba's Qoder Adds Linux ARM64 and LoongArch64 Builds, Plus UOS and Kylin Desktop Support Qoder, Alibaba, LoongArch, Loongson, Linux ARM64, UOS, Kylin, agentic coding https://tpsreport.news/news/moss-image-5b-text-to-image-layout-control TPS — Tokens Per Second en 2026-10-23T07:20:47.366Z MOSI and OpenMOSS open-source MOSS-Image, a 5B text-to-image model for dense text and JSON layout control MOSS-Image, MOSI Intelligence, OpenMOSS, text-to-image, open-source, Hugging Face, layout control https://tpsreport.news/news/chatgpt-chrome-extension-multi-tab-chatgpt-work TPS — Tokens Per Second en 2026-10-23T05:24:21.378Z OpenAI upgrades ChatGPT Chrome extension with multi-tab awareness and ChatGPT Work site control OpenAI, ChatGPT, Chrome, ChatGPT Work, browser extension, AI agents, Gemini https://tpsreport.news/news/anthropic-claude-attack-surface-research-openssh TPS — Tokens Per Second en 2026-10-23T05:22:01.262Z Anthropic: Claude found 493 attack-surface issues across 25 open-source projects, led by OpenSSH fixes Anthropic, Claude, security, OpenSSH, attack surface, Frontier Red Team, OSS Scanner, research https://tpsreport.news/news/liquid-ai-ripl-real-time-inference-engine-lfm TPS — Tokens Per Second en 2026-10-23T05:20:26.562Z Liquid AI launches RIPL, a real-time inference engine for LFM models, with 14.4 to 28.4x lower p95 latency Liquid AI, RIPL, LFM, LFM2.5, inference, latency, vLLM, Qwen3.5 https://tpsreport.news/news/gradium-phonon-31m-on-device-tts-apache-2 TPS — Tokens Per Second en 2026-10-23T05:20:10.424Z Gradium open-sources Phonon 31m, a 31M-parameter on-device TTS model under Apache 2.0 Gradium, Phonon, text-to-speech, on-device, open-source, Apache 2.0 https://tpsreport.news/news/cohere-rerank-5-context-focus-32k-context TPS — Tokens Per Second en 2026-10-23T05:19:53.155Z Cohere Rerank 5 launches in Pro and Fast tiers with 32K context; Pro scores 82.2 on BRIGHT Cohere, Rerank 5, reranking, retrieval, BRIGHT, RAG https://tpsreport.news/news/perplexity-nvidia-jolt-joint-teacher-student-rl TPS — Tokens Per Second en 2026-10-23T05:19:14.306Z Perplexity and NVIDIA's JOLT matches GRPO accuracy on MATH with 6.5x fewer training tokens JOLT, Perplexity, NVIDIA, reinforcement learning, post-training, GRPO, MATH https://tpsreport.news/news/cactus-whistle-16-9-mb-speech-model-vs-whisper-base-cpu TPS — Tokens Per Second en 2026-10-23T05:18:58.498Z Guide benchmarks Cactus Whistle, a 16.9 MB speech model, against Whisper base on a CPU Cactus Whistle, Cactus Compute, Whisper base, speech-to-text, word error rate, CPU inference, benchmark https://tpsreport.news/news/carbon-x-hugging-face-dna-chat-model-14b TPS — Tokens Per Second en 2026-10-23T03:07:42.213Z Hugging Face releases Carbon-X, a 14B-parameter open model that answers natural-language questions about DNA Carbon-X, Hugging Face, genomics, open model, biology AI, DNA, local inference https://tpsreport.news/news/anthropic-claude-formal-proofs-123-libraries TPS — Tokens Per Second en 2026-10-23T03:06:55.075Z Anthropic Says Claude Proved Security Properties for 239 Entry Points Across 123 Libraries Anthropic, Claude, formal verification, security, RISC-V, Theorem, fuzz testing https://tpsreport.news/news/ant-digital-lingdt-decision-1-0-4b-27b TPS — Tokens Per Second en 2026-10-23T02:21:26.569Z Ant Digital Technologies releases LingDT-Decision-1.0 in 4B and 27B sizes for direct structured decisions LingDT-Decision-1.0, Ant Digital Technologies, enterprise AI, decision model, 4B, 27B https://tpsreport.news/news/iflytek-spark-x2-6-293b-a30b-moe-domestic-compute TPS — Tokens Per Second en 2026-10-23T02:21:09.432Z iFLYTEK releases Spark X2.6, a 293B-A30B MoE model trained on domestic compute iFLYTEK, Spark X2.6, MoE, open-source, on-device AI, SparkAudio, China AI https://tpsreport.news/news/nvidia-cosmos-world-model-training-tools-data-factory TPS — Tokens Per Second en 2026-10-23T02:20:47.151Z NVIDIA expands Cosmos with world model training tools, simulation checkpoints and Data Factory blueprints NVIDIA, Cosmos, world models, physical AI, open models, Cosmos Data Factory https://tpsreport.news/news/ling-3-0-flash-sante-healthcare-124b-moe TPS — Tokens Per Second en 2026-10-23T02:07:58.491Z inclusionAI releases Ling-3.0-flash-Sante, a 124B healthcare model with 5.1B active parameters Ling-3.0-flash-Sante, inclusionAI, healthcare AI, open weights, mixture of experts, Hugging Face, medical AI https://tpsreport.news/news/ollama-v0-40-9-phonon-speech-d3-decision-models TPS — Tokens Per Second en 2026-10-23T02:07:29.768Z Ollama v0.40.9 adds Phonon speech generation on Apple Silicon and d3 decision models ollama, phonon, speech, text-to-speech, d3, vllm-semantic-router, apple-silicon, release-notes https://tpsreport.news/news/xiaomi-retires-mimo-v2-5-api-moves-to-mimo-v2-6 TPS — Tokens Per Second en 2026-10-23T02:06:23.169Z Xiaomi retires MiMo-V2.5-Pro and MiMo-V2.5 API models, moves service to MiMo-V2.6 family Xiaomi, MiMo, MiMo-V2.6, MiMo-V2.5, API, deprecation, multimodal https://tpsreport.news/news/openai-decisions-api-widely-available-luna-latency TPS — Tokens Per Second en 2026-10-23T02:06:04.210Z OpenAI opens Decisions API to all customers, cutting latency to 100-180 ms on a reconfigured ChatGPT Luna OpenAI, Decisions API, ChatGPT Luna, TypeSafe, Jev, decision models, latency, OpenRouter https://tpsreport.news/news/qianjue-polibrain-predictive-world-model-cross-embodiment TPS — Tokens Per Second en 2026-10-23T02:05:31.112Z Qianjue launches Polibrain, a world model that predicts task-relevant state changes instead of video frames Qianjue, Polibrain, world model, robotics, embodied AI, cross-embodiment https://tpsreport.news/news/memtensor-memt-1-0-flash-nano-memory-native TPS — Tokens Per Second en 2026-10-23T02:05:14.066Z MemTensor open-sources MemT 1.0 Flash and Nano, models with built-in trainable memory state MemTensor, MemT 1.0, open source, memory, long context, Flash, Nano https://tpsreport.news/news/xiaomi-lens-0-4b-pixel-mask-grounding-lensbench TPS — Tokens Per Second en 2026-10-23T02:04:35.840Z Xiaomi open-sources Xiaomi-Lens-0, a 4B multimodal model that outputs pixel-level mask tokens mid-answer Xiaomi, Xiaomi-Lens-0, SeerRay-Lab, open-source, multimodal, LensBench, Qwen3.5-4B, grounding https://tpsreport.news/news/sensetime-sensenova-roborsi-open-source-27b-embodied-model TPS — Tokens Per Second en 2026-10-23T02:03:52.877Z SenseTime open-sources SenseNova-RoboRSI harness and a 27B embodied model SenseTime, SenseNova-RoboRSI, embodied AI, robotics, open source, GPT-6-Astra https://tpsreport.news/news/xing4-0-29b-intent-china-telecom-open-source TPS — Tokens Per Second en 2026-10-23T02:03:19.121Z China Telecom open-sources Xing4.0-29B-Intent, a 29B intent model, with 2.753M-sample training corpus Xing4.0-29B-Intent, China Telecom, open source, intent recognition, information extraction, structured output, function calling, Apache 2.0 https://tpsreport.news/news/motubrain-2-0-physical-ai-training-livestream TPS — Tokens Per Second en 2026-10-23T02:02:28.995Z Shengshu Technology livestreams full Motubrain 2.0 Physical AI training for about half a month Motubrain, Shengshu Technology, Physical AI, embodied AI, robotics, open source, world model, RSI https://tpsreport.news/news/thinkingbox-training-qwen3-8-27b-grpo-database-rewards TPS — Tokens Per Second en 2026-10-22T22:35:52.634Z Microsoft's ThinkingBox-Training lifts Qwen3.8-27B fully reliable task completions from 38 to 107 of 507 ThinkingBox, Microsoft, Hugging Face, Qwen3.8-27B, GRPO, reinforcement learning, agents, benchmark https://tpsreport.news/news/canva-craft-in-house-image-model-2k-two-stage TPS — Tokens Per Second en 2026-10-22T22:05:49.620Z Canva launches Craft, its first in-house image model, with two-stage design planning and 2K output Canva, Canva Craft, image generation, Design Arena, Model Context Protocol, OpenAI, GPT-Image-2.5, AI design https://tpsreport.news/news/replit-monitoring-agent-autonomous-production-fixes TPS — Tokens Per Second en 2026-10-22T20:35:48.382Z Replit Monitoring Agent investigates production alerts, validates fixes, and can auto-deploy them replit, monitoring-agent, ai-agents, autofix, production-monitoring, devops, app-publishing https://tpsreport.news/news/gemini-personal-intelligence-google-maps-saved-places-deep-think-argon TPS — Tokens Per Second en 2026-10-22T19:35:46.437Z Gemini app's Personal Intelligence now reads Google Maps searches and saved places; Argon to replace Deep Think Gemini, Google, Personal Intelligence, Google Maps, Deep Think, Gemini 4 Argon, Gemini 3.1 Pro, Android Auto https://tpsreport.news/news/microsoft-mai-dxo-83-percent-complex-diagnosis-science-study TPS — Tokens Per Second en 2026-10-22T18:50:49.934Z Microsoft's MAI-DxO scores 83% vs 36% for physicians on complex NEJM cases in Science study Microsoft, MAI-DxO, medical AI, Mustafa Suleyman, healthcare, benchmark, GPT-5.5, medical superintelligence https://tpsreport.news/news/amazon-polly-generative-engine-upgrade-14-voices-hindi-finnish TPS — Tokens Per Second en 2026-10-22T18:20:53.682Z Amazon Polly Generative engine upgrade adds 14 voices, first Hindi and Finnish support, at unchanged pricing Amazon Polly, AWS, text-to-speech, generative voices, Hindi, Finnish, voice agents, GovCloud https://tpsreport.news/news/olmoearth-v1-3-satellite-embeddings TPS — Tokens Per Second en 2026-10-22T16:20:54.826Z Ai2 releases OlmoEarth v1.3 satellite embedding model with open global 10m embeddings OlmoEarth, Ai2, satellite embeddings, geospatial AI, open source, remote sensing, foundation models, Sentinel-2 https://tpsreport.news/news/openai-gpt-rosalind-agents-mouse-brain-crispr-atlas-32-minutes TPS — Tokens Per Second en 2026-10-22T10:20:54.970Z OpenAI agents on GPT-Rosalind analyze mouse brain CRISPR atlas in 32 minutes, versus months by hand OpenAI, GPT-Rosalind, AI agents, biology, CRISPR, Scripps Research, PerturbAI, bioRxiv https://tpsreport.news/news/ibm-granite-timeseries-ensemble-r1-gift-eval TPS — Tokens Per Second en 2026-10-22T08:05:51.516Z IBM's Granite Timeseries Ensemble pools four open models and edges each member on GIFT-Eval point-forecast MASE IBM, Granite, time-series, forecasting, GIFT-Eval, ensemble, open-source, Hugging Face https://tpsreport.news/news/sage-a1-5-35b-moe-agentic-model TPS — Tokens Per Second en 2026-10-22T01:51:07.158Z Shanghai AI Lab releases Sage-A1.5, a 35B MoE agentic model with 256K native context Sage-A1.5, Shanghai AI Laboratory, InternScience, open weights, Mixture-of-Experts, agentic AI, GAIA, SWE-bench Pro https://tpsreport.news/news/recursive-flint-40b-moe-laptop-coding-model TPS — Tokens Per Second en 2026-10-21T20:35:56.660Z Recursive releases Flint, a 40B-parameter open coding model that runs on laptops under Apache 2.0 Flint, Recursive, open-weight, coding model, mixture-of-experts, Apache 2.0, on-device AI, Richard Socher https://tpsreport.news/news/anthropic-claude-duet-always-on-agent-research-preview TPS — Tokens Per Second en 2026-10-21T19:50:46.189Z Anthropic launches Claude Duet, an always-on agent, in research preview for US Pro and Max users Anthropic, Claude Duet, Claude Opus 5.5, AI agents, product launch, Claude, research preview https://tpsreport.news/news/nvidia-kumo-relational-foundation-model-relational-data TPS — Tokens Per Second en 2026-10-21T18:35:53.538Z NVIDIA Kumo Relational predicts from databases using up to 10,000 in-context examples, no training NVIDIA, Kumo Relational, foundation model, relational data, tabular ML, in-context learning, NVIDIA NIM, RelBench https://tpsreport.news/news/google-cloud-openai-gpt-6-private-preview TPS — Tokens Per Second en 2026-10-21T17:50:57.860Z Google Cloud adds three OpenAI GPT-6 models in private preview, alongside Gemini and Claude Google Cloud, OpenAI, GPT-6, Gemini, Claude, Enterprise AI, Cloud Platforms, Private Preview