https://aiinsiders.net/article/a-proxy-lets-codex-run-any-model-not-just-openais AI Insiders en 2026-07-23T09:00:00.000Z A Proxy Lets Codex Run Any Model, Not Just OpenAI's https://aiinsiders.net/article/a-tokenizer-claims-1000x-speed-over-hugging-face-by-its-own AI Insiders en 2026-07-23T09:00:00.000Z A tokenizer claims 1,000x speed over Hugging Face, by its own math https://aiinsiders.net/article/acp-v2-draft-rewrites-how-coding-agents-talk-to-editors AI Insiders en 2026-07-23T09:00:00.000Z ACP v2 Draft Rewrites How Coding Agents Talk to Editors https://aiinsiders.net/article/claude-code-desktop-can-now-watch-its-own-ios-app-run AI Insiders en 2026-07-23T09:00:00.000Z Claude Code Desktop Can Now Watch Its Own iOS App Run https://aiinsiders.net/article/cognition-lets-devin-run-on-your-own-hardware-not-its-cloud AI Insiders en 2026-07-23T09:00:00.000Z Cognition Lets Devin Run on Your Own Hardware, Not Its Cloud https://aiinsiders.net/article/google-says-pretraining-has-begun-for-gemini-4 AI Insiders en 2026-07-23T09:00:00.000Z Google says pretraining has begun for Gemini 4 https://aiinsiders.net/article/google-splits-gemini-flash-into-three-models-for-speed-cost AI Insiders en 2026-07-23T09:00:00.000Z Google Splits Gemini Flash Into Three Models for Speed, Cost, Security https://aiinsiders.net/article/metas-sam-3-and-dinov3-now-run-inside-a-doe-national-lab AI Insiders en 2026-07-23T09:00:00.000Z Meta's SAM 3 and DINOv3 Now Run Inside a DOE National Lab Pipeline https://aiinsiders.net/article/microsoft-ships-mage-a-4b-image-model-line-built-for-labs AI Insiders en 2026-07-23T09:00:00.000Z Microsoft Ships Mage, a 4B Image Model Line Built for Labs, Not Products https://aiinsiders.net/article/openai-model-exploited-a-zero-day-to-raid-hugging-face-for AI Insiders en 2026-07-23T09:00:00.000Z OpenAI model exploited a zero-day to raid Hugging Face for answers https://aiinsiders.net/article/poolside-ships-laguna-s-21-a-118b-coding-model-at-8b-cost AI Insiders en 2026-07-23T09:00:00.000Z Poolside Ships Laguna S 2.1, a 118B Coding Model at 8B Cost https://aiinsiders.net/article/qwens-new-image-model-targets-slide-decks-not-just-pretty AI Insiders en 2026-07-23T09:00:00.000Z Qwen's new image model targets slide decks, not just pretty pictures https://aiinsiders.net/article/the-five-layers-a-real-web-agent-needs-past-chromium AI Insiders en 2026-07-23T09:00:00.000Z The five layers a real web agent needs past Chromium https://aiinsiders.net/article/the-real-bottleneck-for-robot-ai-isnt-the-model-its-the AI Insiders en 2026-07-23T09:00:00.000Z The Real Bottleneck for Robot AI Isn't the Model. It's the Simulator https://aiinsiders.net/article/why-distillation-makes-the-frontier-ai-moat-so-shallow AI Insiders en 2026-07-23T09:00:00.000Z Why Distillation Makes the Frontier AI Moat So Shallow https://aiinsiders.net/article/zvi-mowshowitz-openais-fixes-manage-symptoms-not AI Insiders en 2026-07-23T09:00:00.000Z Zvi Mowshowitz: OpenAI's Fixes Manage Symptoms, Not Misalignment https://aiinsiders.net/edition/2026-07-23 AI Insiders en 2026-07-23T00:00:00.000Z The Sandbox Broke Twice, and the Fix Still Isn't the Fix https://aiinsiders.net/article/ai-is-cutting-drug-development-costs-by-70percent-approvals AI Insiders en 2026-07-22T09:00:00.000Z AI Is Cutting Drug Development Costs by 70%. Approvals Are Still Zero https://aiinsiders.net/article/amds-helios-rack-system-lands-microsoft-challenges-nvidia AI Insiders en 2026-07-22T09:00:00.000Z AMD's Helios rack system lands Microsoft, challenges Nvidia grip https://aiinsiders.net/article/anthropic-shuts-down-conway-agent-test-sets-july-24-data AI Insiders en 2026-07-22T09:00:00.000Z Anthropic Shuts Down Conway Agent Test, Sets July 24 Data Deadline https://aiinsiders.net/article/cognition-acquires-tierzero-to-bring-incident-automation-to AI Insiders en 2026-07-22T09:00:00.000Z Cognition Acquires TierZero to Bring Incident Automation to Devin https://aiinsiders.net/article/cursors-new-coding-swarm-cuts-ai-agent-costs-by-87-percent AI Insiders en 2026-07-22T09:00:00.000Z Cursor's New Coding Swarm Cuts AI Agent Costs by 87 Percent https://aiinsiders.net/article/google-reportedly-building-new-chip-to-cut-ai-power-costs AI Insiders en 2026-07-22T09:00:00.000Z Google Reportedly Building New Chip to Cut AI Power Costs https://aiinsiders.net/article/kimi-k3-shows-open-labs-are-trading-compute-for-storage AI Insiders en 2026-07-22T09:00:00.000Z Kimi K3 Shows Open Labs Are Trading Compute for Storage https://aiinsiders.net/article/kimi-k3s-28t-model-cuts-chinas-ai-gap-to-months-not-years AI Insiders en 2026-07-22T09:00:00.000Z Kimi K3's 2.8T Model Cuts China's AI Gap to Months, Not Years https://aiinsiders.net/article/moonshot-ai-launches-kimi-work-a-desktop-agent-for-files AI Insiders en 2026-07-22T09:00:00.000Z Moonshot AI launches Kimi Work, a desktop agent for files and browsers https://aiinsiders.net/article/nvidia-shrinks-cosmos-3-to-run-its-world-model-on-the-robot AI Insiders en 2026-07-22T09:00:00.000Z NVIDIA shrinks Cosmos 3 to run its world model on the robot itself https://aiinsiders.net/article/openais-internal-agent-broke-its-sandbox-to-post-on-github AI Insiders en 2026-07-22T09:00:00.000Z OpenAI's Internal Agent Broke Its Sandbox to Post on GitHub https://aiinsiders.net/article/ramps-llm-gateway-cuts-ai-costs-26percent-with-failure AI Insiders en 2026-07-22T09:00:00.000Z Ramp's LLM Gateway Cuts AI Costs 26% With Failure-Aware Routing https://aiinsiders.net/article/researcher-argues-the-harness-not-the-model-drives AI Insiders en 2026-07-22T09:00:00.000Z Researcher argues the harness, not the model, drives generalization https://aiinsiders.net/article/sushanth-raman-debuts-supply-chain-custom-models-on-x AI Insiders en 2026-07-22T09:00:00.000Z Sushanth Raman debuts supply chain Custom Models on X https://aiinsiders.net/article/tokens-got-cheaper-your-ai-bill-did-not-follow AI Insiders en 2026-07-22T09:00:00.000Z Tokens Got Cheaper. Your AI Bill Did Not Follow. https://aiinsiders.net/article/xiaomis-new-robot-model-trains-mostly-on-human-video-not AI Insiders en 2026-07-22T09:00:00.000Z Xiaomi's New Robot Model Trains Mostly on Human Video, Not Robots https://aiinsiders.net/article/zai-brings-a-1-gigawatt-all-chinese-chip-data-center-online AI Insiders en 2026-07-22T09:00:00.000Z Z.ai Brings a 1-Gigawatt, All-Chinese-Chip Data Center Online https://aiinsiders.net/edition/2026-07-22 AI Insiders en 2026-07-22T00:00:00.000Z China Closes the Gap, Agents Test Their Limits https://aiinsiders.net/article/a-dollar400-million-loan-bets-non-nvidia-chips-hold-their AI Insiders en 2026-07-21T12:00:00.000Z A $400 Million Loan Bets Non-Nvidia Chips Hold Their Value https://aiinsiders.net/article/alibaba-open-sources-zhenwu-chip-software-to-crack-cudas AI Insiders en 2026-07-21T12:00:00.000Z Alibaba Open-Sources Zhenwu Chip Software to Crack CUDA's Grip https://aiinsiders.net/article/alibabas-qwen38-joins-the-2-trillion-parameter-open-weight AI Insiders en 2026-07-21T12:00:00.000Z Alibaba's Qwen3.8 Joins the 2-Trillion-Parameter Open Weight Sprint https://aiinsiders.net/article/apple-widens-openai-trade-secret-fight-with-preservation AI Insiders en 2026-07-21T12:00:00.000Z Apple Widens OpenAI Trade Secret Fight With Preservation Letters https://aiinsiders.net/article/claude-code-and-codex-handle-goal-differently-and-neither AI Insiders en 2026-07-21T12:00:00.000Z Claude Code and Codex Handle /goal Differently, and Neither Is Great https://aiinsiders.net/article/claude-fable-5-joins-max-team-premium-plan-limits-july-20 AI Insiders en 2026-07-21T12:00:00.000Z Claude Fable 5 joins Max, Team Premium plan limits July 20 https://aiinsiders.net/article/google-deepmind-researcher-quits-over-unrestricted-military AI Insiders en 2026-07-21T12:00:00.000Z Google DeepMind Researcher Quits Over Unrestricted Military AI Deal https://aiinsiders.net/article/google-tests-gemini-live-on-web-portable-skills-for-all AI Insiders en 2026-07-21T12:00:00.000Z Google Tests Gemini Live on Web, Portable Skills for All Chats https://aiinsiders.net/article/moonshot-ai-targets-hong-kong-ipo-at-dollar30-billion AI Insiders en 2026-07-21T12:00:00.000Z Moonshot AI Targets Hong Kong IPO at $30 Billion Valuation https://aiinsiders.net/article/moonshot-pauses-new-kimi-k3-signups-as-demand-outstrips-gpus AI Insiders en 2026-07-21T12:00:00.000Z Moonshot Pauses New Kimi K3 Signups as Demand Outstrips GPUs https://aiinsiders.net/article/moonshot-ships-kimi-code-a-terminal-agent-that-mirrors AI Insiders en 2026-07-21T12:00:00.000Z Moonshot ships Kimi Code, a terminal agent that mirrors Claude Code https://aiinsiders.net/article/netflix-runs-its-own-llm-inference-stack-instead-of-buying AI Insiders en 2026-07-21T12:00:00.000Z Netflix Runs Its Own LLM Inference Stack Instead of Buying APIs https://aiinsiders.net/article/quesmas-fix-for-ai-research-costs-cheap-models-first-deep AI Insiders en 2026-07-21T12:00:00.000Z Quesma's Fix for AI Research Costs: Cheap Models First, Deep Research Last https://aiinsiders.net/article/sakana-ai-trains-neural-nets-without-breaking-a-core-brain AI Insiders en 2026-07-21T12:00:00.000Z Sakana AI trains neural nets without breaking a core brain rule https://aiinsiders.net/article/zai-nears-dollar1bn-in-sales-by-giving-its-best-models-away AI Insiders en 2026-07-21T12:00:00.000Z Z.ai nears $1bn in sales by giving its best models away free https://aiinsiders.net/edition/2026-07-21 AI Insiders en 2026-07-21T00:00:00.000Z China's Open Weight Surge Meets Western Legal Heat