https://tokenstrom.com/posts/nvidia-vp-says-letter-demanding-jensen-huang-s-return-is-obviously-fak/Tokenstromen2026-11-14T13:22:07+00:00Nvidia VP says letter demanding Jensen Huang's return is 'obviously fake' https://tokenstrom.com/posts/agents-vs-cuda-much-faster-at-reading-prompts-barely-faster-at-answeri/Tokenstromen2026-11-14T12:59:49+00:00Agents vs. CUDA: Much Faster at Reading Prompts, Barely Faster at Answering https://tokenstrom.com/posts/proofs-yes-pipettes-no-why-ai-scientists-struggle-in-real-labs/Tokenstromen2026-11-14T12:49:42+00:00Proofs, Yes. Pipettes, No: Why AI Scientists Struggle in Real Labs https://tokenstrom.com/posts/plenio-2-0-turns-comfyui-into-a-local-studio-from-song-idea-to-finishe/Tokenstromen2026-11-14T12:42:36+00:00Plenio 2.0 turns ComfyUI into a local studio, from song idea to finished master https://tokenstrom.com/posts/one-developer-a-fleet-of-ai-agents-free-cad-app-solidcraft-tops-1-5m-d/Tokenstromen2026-11-14T12:29:05+00:00One developer, a fleet of AI agents: free CAD app SolidCraft tops 1.5M downloads https://tokenstrom.com/posts/ai-models-know-when-they-re-being-tested-what-are-safety-evals-still-w/Tokenstromen2026-11-14T10:55:14+00:00AI models know when they're being tested. What are safety evals still worth? https://tokenstrom.com/posts/agent-browsers-fact-checked-what-10x-faster-leaves-out/Tokenstromen2026-11-14T10:47:53+00:00Agent browsers fact-checked: what "10x faster" leaves out https://tokenstrom.com/posts/hackers-turn-ai-agents-loose-on-chipmakers-and-defense-firms-breaching/Tokenstromen2026-11-14T10:29:28+00:00Hackers turn AI agents loose on chipmakers and defense firms, breaching ten targets https://tokenstrom.com/posts/secret-two-letter-code-makes-a-kitchen-robot-point-its-knife-at-a-hand/Tokenstromen2026-11-14T10:10:50+00:00Secret two-letter code makes a kitchen robot point its knife at a hand, in the lab https://tokenstrom.com/posts/when-your-chatbot-shows-you-buttons-who-decides-what-confirm-actually/Tokenstromen2026-11-14T08:53:27+00:00When your chatbot shows you buttons, who decides what “Confirm” actually does? https://tokenstrom.com/posts/who-controls-the-ai-engine-llama-cpp-sglang-and-vllm-take-three-differ/Tokenstromen2026-11-14T08:45:55+00:00Who controls the AI engine? llama.cpp, SGLang and vLLM take three different paths https://tokenstrom.com/posts/remove-bg-is-shutting-down-7-free-local-background-removers-license-ch/Tokenstromen2026-11-14T06:51:05+00:00remove.bg is shutting down: 7 free local background removers, license-checked https://tokenstrom.com/posts/gpt-6-astra-beats-robot-ais-but-only-while-the-simulation-stands-still/Tokenstromen2026-11-14T06:42:33+00:00GPT-6 Astra beats robot AIs, but only while the simulation stands still https://tokenstrom.com/posts/claude-code-projects-one-ai-now-runs-a-whole-team-of-parallel-ai-worke/Tokenstromen2026-11-14T04:35:50+00:00Claude Code Projects: One AI now runs a whole team of parallel AI workers https://tokenstrom.com/posts/google-s-ai-agent-teams-crack-decades-old-math-problems-and-rewrite-a/Tokenstromen2026-11-14T02:00:33+00:00Google's AI agent teams crack decades-old math problems and rewrite a kernel in Rust https://tokenstrom.com/posts/portal2bench-only-gpt-6-astra-cracks-the-portal-2-chamber-that-stumps/Tokenstromen2026-11-13T23:41:54+00:00Portal2Bench: Only GPT-6 Astra cracks the Portal 2 chamber that stumps every other model https://tokenstrom.com/posts/llama-cpp-hits-v1-the-backbone-of-local-ai-is-officially-stable/Tokenstromen2026-11-13T20:28:52+00:00llama.cpp hits v1: the backbone of local AI is officially stable https://tokenstrom.com/posts/lingbot-touch-tiny-open-source-model-gives-robot-hands-a-sense-of-touc/Tokenstromen2026-11-13T18:26:59+00:00LingBot-Touch: Tiny open-source model gives robot hands a sense of touch https://tokenstrom.com/posts/mimvox-clones-your-voice-locally-speaks-60-languages-and-redubs-your-v/Tokenstromen2026-11-13T18:09:15+00:00Mimvox clones your voice locally, speaks 60 languages and redubs your videos https://tokenstrom.com/posts/tontaube-v1-1-open-tts-with-voice-cloning-german-as-a-core-language-an/Tokenstromen2026-11-13T17:33:52+00:00Tontaube V1.1: Open TTS with voice cloning, German as a core language, and CPU support https://tokenstrom.com/posts/suspect-freed-over-ai-police-report-what-draft-one-and-truleo-really-d/Tokenstromen2026-11-13T16:51:02+00:00Suspect Freed Over AI Police Report: What Draft One and Truleo Really Deliver https://tokenstrom.com/posts/stock-ai-voice-real-voice-actor-why-you-re-liable-not-the-provider/Tokenstromen2026-11-13T16:44:59+00:00Stock AI Voice, Real Voice Actor: Why You're Liable, Not the Provider https://tokenstrom.com/posts/anthropic-s-attack-ai-anvil-breaks-safety-filters-faster-than-human-re/Tokenstromen2026-11-13T16:34:31+00:00Anthropic's attack AI Anvil breaks safety filters faster than human red teamers https://tokenstrom.com/posts/firmware-hacking-with-claude-what-german-and-eu-law-actually-allows/Tokenstromen2026-11-13T14:55:36+00:00Firmware hacking with Claude: What German and EU law actually allows https://tokenstrom.com/posts/ai-weather-beats-physics-but-only-when-ai-data-keeps-score/Tokenstromen2026-11-13T14:43:20+00:00AI weather "beats physics" – but only when AI data keeps score? https://tokenstrom.com/posts/perplexity-decider-v1-3-open-weight-models-that-make-decisions-instead/Tokenstromen2026-11-13T13:31:26+00:00Perplexity Decider v1.3: Open-weight models that make decisions instead of writing text https://tokenstrom.com/posts/the-leak-beneath-the-ai-cloud-why-inference-software-is-spilling-your/Tokenstromen2026-11-13T13:06:49+00:00The leak beneath the AI cloud: why inference software is spilling your data https://tokenstrom.com/posts/ai-agents-in-your-brokerage-where-claude-can-trade-and-where-the-brake/Tokenstromen2026-11-13T12:58:47+00:00AI agents in your brokerage: where Claude can trade, and where the brakes are missing https://tokenstrom.com/posts/haru-4b-open-japanese-voice-model-that-says-mhm-and-handles-interrupti/Tokenstromen2026-11-13T12:34:07+00:00Haru 4B: Open Japanese voice model that says "mhm" and handles interruptions https://tokenstrom.com/posts/lara-mini-europe-s-45-language-model-outscores-bigger-rivals-in-its-cl/Tokenstromen2026-11-13T11:56:35+00:00Lara Mini: Europe's 45-language model outscores bigger rivals in its class https://tokenstrom.com/posts/ten-lines-of-ai-code-why-the-games-purity-test-breaks-down-at-the-engi/Tokenstromen2026-11-13T11:01:59+00:00Ten Lines of AI Code: Why the Games 'Purity Test' Breaks Down at the Engine https://tokenstrom.com/posts/after-the-mittwald-hack-how-nine-hosts-secure-their-ai-interfaces/Tokenstromen2026-11-13T10:52:25+00:00After the Mittwald Hack: How Nine Hosts Secure Their AI Interfaces https://tokenstrom.com/posts/world-labs-opens-up-atlas-turn-a-sketch-or-a-sentence-into-a-3d-world/Tokenstromen2026-11-13T10:32:54+00:00World Labs opens up Atlas: turn a sketch or a sentence into a 3D world you can explore https://tokenstrom.com/posts/german-speech-recognition-tiny-parakeet-keeps-pace-with-whisper-turbo/Tokenstromen2026-11-13T08:49:56+00:00German speech recognition: tiny Parakeet keeps pace with Whisper, Turbo falls behind https://tokenstrom.com/posts/russia-s-ai-vpn-filter-what-it-sees-without-decrypting-and-where-it-mi/Tokenstromen2026-11-13T08:41:58+00:00Russia's AI VPN filter: what it sees without decrypting, and where it misfires https://tokenstrom.com/posts/hackers-put-an-ai-agent-to-work-as-a-spy-then-got-caught-by-their-own/Tokenstromen2026-11-13T08:29:44+00:00Hackers put an AI agent to work as a spy, then got caught by their own open server https://tokenstrom.com/posts/command-or-just-text-ibm-s-llmon-wants-to-stop-prompt-injection-with-g/Tokenstromen2026-11-13T06:49:08+00:00Command or just text? IBM's LLMON wants to stop prompt injection with grammar https://tokenstrom.com/posts/black-hole-holds-up-maya-pyramid-doesn-t-how-to-vet-ai-discoveries/Tokenstromen2026-11-13T06:40:44+00:00Black hole holds up, Maya pyramid doesn't: how to vet AI discoveries https://tokenstrom.com/posts/bring-back-the-boss-nvidia-staff-push-for-jensen-huang-s-return/Tokenstromen2026-11-13T04:07:55+00:00'Bring back the boss': Nvidia staff push for Jensen Huang's return https://tokenstrom.com/posts/scanruler-free-browser-tool-measures-3d-scans-and-an-ai-agent-rebuilds/Tokenstromen2026-11-13T02:20:58+00:00ScanRuler: Free browser tool measures 3D scans, and an AI agent rebuilds them in CAD https://tokenstrom.com/posts/edge0-145b-an-open-145-billion-parameter-model-that-s-meant-to-run-on/Tokenstromen2026-11-13T01:29:32+00:00Edge0-145B: An open 145-billion-parameter model that's meant to run on your laptop https://tokenstrom.com/posts/pixel-phones-describe-the-widget-you-want-and-gemini-builds-it-for-you/Tokenstromen2026-11-13T00:31:57+00:00Pixel phones: Describe the widget you want, and Gemini builds it for you https://tokenstrom.com/posts/openai-teases-its-next-hacking-ai-77-percent-on-an-exploit-benchmark/Tokenstromen2026-11-13T00:21:49+00:00OpenAI teases its next hacking AI: 77 percent on an exploit benchmark https://tokenstrom.com/posts/a-cancer-vaccine-for-one-father-popvax-puts-tumor-data-and-design-on-g/Tokenstromen2026-11-13T00:20:02+00:00A cancer vaccine for one father: PopVax puts tumor data and design on GitHub https://tokenstrom.com/posts/thinking-machines-releases-inkling-2-small-weights-a-594b-moe-under-ap/Tokenstromen2026-11-12T22:28:52+00:00Thinking Machines releases Inkling 2 Small weights: a 594B MoE under Apache 2.0 https://tokenstrom.com/posts/a-16b-model-on-a-cheap-tv-box-smolmaxxing-squeezes-ai-out-of-ancient-h/Tokenstromen2026-11-12T20:49:56+00:00A 16B model on a cheap TV box: smolmaxxing squeezes AI out of ancient hardware https://tokenstrom.com/posts/half-of-it-is-slop-ai-patch-flood-drives-a-linux-kernel-developer-out/Tokenstromen2026-11-12T18:52:07+00:00'Half of it is slop': AI patch flood drives a Linux kernel developer out of code review https://tokenstrom.com/posts/built-with-gemini-4-carbon-google-exec-gets-windows-11-running-on-the/Tokenstromen2026-11-12T18:51:09+00:00Built with Gemini 4 Carbon: Google exec gets Windows 11 running on the iPhone Duo https://tokenstrom.com/posts/salesforce-s-sov-transcribe-1-is-now-the-best-open-speech-recognition/Tokenstromen2026-11-12T16:59:33+00:00Salesforce's SOV-Transcribe-1 is now the best open speech recognition model https://tokenstrom.com/posts/ai-agents-took-down-a-wind-farm-in-66-minutes-for-20/Tokenstromen2026-11-12T16:56:33+00:00AI agents took down a wind farm in 66 minutes – for $20 https://tokenstrom.com/posts/gemini-refused-the-image-then-the-download-button-handed-it-over-anywa/Tokenstromen2026-11-12T15:29:15+00:00Gemini refused the image – then the download button handed it over anyway https://tokenstrom.com/posts/samsung-and-google-launch-30-gemini-powered-ai-glasses-starting-at-395/Tokenstromen2026-11-12T15:24:41+00:00Samsung and Google launch 30 Gemini-powered AI glasses, starting at $395 https://tokenstrom.com/posts/jensen-huang-quits-as-nvidia-ceo-amid-affair-and-espionage-scandal/Tokenstromen2026-11-12T15:24:41+00:00Jensen Huang quits as Nvidia CEO amid affair and espionage scandal https://tokenstrom.com/posts/kv-cache-on-a-diet-what-learned-bouncers-forget-in-long-contexts/Tokenstromen2026-11-12T14:10:33+00:00KV cache on a diet: what learned “bouncers” forget in long contexts https://tokenstrom.com/posts/everyone-beats-alphafold-3-just-never-on-the-same-test-what-drug-ai-re/Tokenstromen2026-11-12T14:02:56+00:00Everyone beats AlphaFold 3, just never on the same test: what drug-AI records show https://tokenstrom.com/posts/pubg-maker-krafton-releases-raon-1-nano-a-small-ai-model-that-runs-on/Tokenstromen2026-11-12T13:54:41+00:00PUBG maker Krafton releases Raon-1 nano, a small AI model that runs on your own PC https://tokenstrom.com/posts/elevenlabs-scribe-v3-speech-recognition-that-stops-cutting-callers-off/Tokenstromen2026-11-12T13:51:05+00:00ElevenLabs Scribe v3: speech recognition that stops cutting callers off mid-sentence