Meta Launches Muse Code With a 92% Discount for Training Rights Agent and coding tools 80 1 /posts/meta-launches-muse-code-with-a-92-discount-for-training-rights Anthropic’s Dario Amodei Says AI Must Deliver, Not Advertise, Its Way Out of a Trust Crisis Safety evaluations policy 30 3 /posts/anthropic-s-dario-amodei-says-ai-must-deliver-not-advertise-its-way-out-of-a-trust-crisis Zuckerberg’s superintelligence promise runs into AI’s trust test Safety evaluations policy 70 1 /posts/zuckerberg-s-superintelligence-promise-runs-into-ai-s-trust-test A Naming Error Let Anthropic Models Reach a Real Production Database Safety evaluations models 90 2 /posts/a-naming-error-let-anthropic-models-reach-a-real-production-database Alibaba Launches HappyShrimp for One-Prompt Songs, but Leaves Key Use Terms Unstated Leaderboards models 80 1 /posts/alibaba-launches-happyshrimp-for-one-prompt-songs-but-leaves-key-use-terms-unstated Zhipu Says GLM-5.3 Can Find Bugs Across an Exploitation Chain Safety evaluations models 80 2 /posts/zhipu-says-glm-5-3-can-find-bugs-across-an-exploitation-chain Alibaba Releases Downloadable Qwen3.8-Max Weights Alongside a Laptop Model Agent and coding models 80 3 /posts/alibaba-puts-qwen-on-laptops-and-opens-its-biggest-model-as-meta-courts-the-same-developers Nvidia’s Reported Lancium Deal Would Put Grid Connections at the Center of Its AI Buildout Leaderboards infrastructure 80 3 /posts/nvidia-s-ai-infrastructure-bet-reaches-the-grid-not-just-the-data-center ByteDance’s Hollywood Copyright Truce Leaves the Guardrails Private Leaderboards policy 70 4 /posts/hollywood-and-bytedance-trade-a-copyright-fight-for-an-unseen-ai-guardrail-deal Snowflake Wants AI Apps to Stop Paying Frontier-Model Prices for Every Task Agent and coding products 80 2 /posts/snowflake-wants-ai-apps-to-stop-paying-frontier-model-prices-for-every-task OpenAI Funds a Study of How AI Could Shift Tax Revenue Safety evaluations policy 72 1 /posts/openai-funds-a-study-of-how-ai-could-shift-tax-revenue OpenAI Paused Deployment-Bound Training as It Tightened Frontier Security Safety evaluations models 80 1 /posts/openai-paused-deployment-bound-training-as-it-tightened-frontier-security Etched’s $700 Million Round Puts a $21 Billion Price on Inference Hardware Capability evaluations startups 90 3 /posts/etched-s-700-million-round-puts-a-21-billion-price-on-inference-hardware Wispr’s $280 Million Bet Is That Voice Can Move Beyond Dictation Leaderboards startups 90 4 /posts/wispr-s-280-million-bet-is-that-voice-can-move-beyond-dictation AgentX Replays Claude Code Sessions to Test AI Serving Systems Agent and coding infrastructure 70 1 /posts/agentx-replays-claude-code-sessions-to-test-ai-serving-systems TestMu’s New Agent Test Shows What It Can’t Verify Safety evaluations tools 80 1 /posts/testmu-s-new-agent-test-shows-what-it-can-t-verify LMCache Reworks the Cache Plumbing That Can Stall Long-Running Agents Agent and coding infrastructure 70 1 /posts/lmcache-reworks-the-cache-plumbing-that-can-stall-long-running-agents Dane County Routes Non-Emergency Calls Through AVA to Protect 911 Capacity Safety evaluations infrastructure 70 4 /posts/dane-county-puts-an-ai-gatekeeper-on-its-non-emergency-line Oakley Buys Majority Stake in Graphwise as It Plans Expansion and Acquisitions Agent and coding business 80 2 /posts/oakley-takes-majority-control-of-graphwise-as-it-plans-a-global-ai-push Jinho Jang Puts a 27B Refusal-Removed Model Into a Local Download Safety evaluations models 80 0 /posts/jinho-jang-puts-a-27b-refusal-removed-model-into-a-local-download Nome Uses AI to Find Rare-Disease Treatment Paths Before the Hard Work Begins Agent and coding startups 60 1 /posts/nome-promises-rare-disease-families-a-treatment-path-the-clinical-work-still-lies-ahead Anthropic Will Watermark Claude Text by Steering Its Word Choices Capability evaluations models 70 1 /posts/anthropic-will-watermark-claude-text-by-steering-its-word-choices A Reporter’s LLM Wiki Speeds Recall. He Still Checks the Original Sources. Agent and coding tools 65 1 /posts/casey-newton-built-an-ai-wiki-for-his-beat-now-he-has-to-keep-it-alive Vivodyne Builds a Human-Tissue Data Factory for Drug AI Capability evaluations startups 70 1 /posts/vivodyne-built-a-robot-lab-for-the-human-biology-data-ai-still-lacks SandboxAQ Opens a Drug-Screening Model That Does Not Need Protein Structures Capability evaluations models 78 0 /posts/sandboxaq-opens-a-drug-screening-model-that-does-not-need-protein-structures OpenAI Tests Cross-Session Safety Checks Without Prompt Access Safety evaluations products 80 0 /posts/openai-tests-cross-session-safety-checks-without-prompt-access OpenAI Takes Codex From Coding to Tax Returns Agent and coding products 70 0 /posts/openai-takes-codex-from-coding-to-tax-returns TrueFoundry Gives Away Its Agent Runtime to Sell the Layer Beneath It Agent and coding tools 80 0 /posts/truefoundry-gives-away-its-agent-runtime-to-sell-the-layer-beneath-it Palomar Opens a Lean Registry That Checks Proofs and Their Descriptions Capability evaluations tools 70 0 /posts/palomar-opens-a-lean-proof-registry-with-an-llm-semantic-check UK Cyber Tests Show AI Agents Going Beyond the Technical Task Safety evaluations models 80 0 /posts/ai-cyber-tests-are-reaching-real-targets-not-just-sandboxes At This D.C. Charter, AI Permission Changes With the Assignment Leaderboards policy 60 2 /posts/this-d-c-charter-made-ai-a-schoolwide-skill-not-a-shortcut Mistral Gives Enterprise AI Five Ways to Keep Digging Through Documents Agent and coding products 80 0 /posts/mistral-gives-enterprise-ai-five-ways-to-keep-digging-through-documents Simple AI Opens 2,000 Hours of Robot Training Data Collected Without Robots Leaderboards models 70 0 /posts/simple-ai-opens-2-000-hours-of-robot-training-data-collected-without-robots Ramp Opens Its AI Router to U.S. Customers, With a One-Year Data Default Capability evaluations products 70 1 /posts/ramp-opens-its-ai-router-to-u-s-customers-with-a-one-year-data-default Anthropic Opens Claude Academy as Free Training for Its AI Products Agent and coding products 70 0 /posts/anthropic-opens-claude-academy-as-free-training-for-its-ai-products Micro1’s Reported $500M Run Rate Tests the Economics of AI Training Data Leaderboards startups 75 1 /posts/micro1-s-reported-500m-run-rate-tests-the-economics-of-ai-training-data Callosum Raises $100 Million as It Routes AI Work Across Models and Chips Capability evaluations startups 80 2 /posts/callosum-raises-100-million-to-route-ai-work-across-models-and-chips Starcloud Has $250 Million for Orbital AI. It Still Needs a Ride. Leaderboards startups 90 1 /posts/starcloud-has-250-million-for-orbital-ai-it-still-needs-a-ride Open Models Are Catching the Frontier Faster. Benchmark Scores Aren’t the Whole Contest. Agent and coding models 80 1 /posts/open-models-are-catching-the-frontier-faster-benchmark-scores-aren-t-the-whole-contest Nvidia Maps AI Memory Between Models Instead of Making Them Start Over Capability evaluations models 80 1 /posts/nvidia-maps-ai-memory-between-models-instead-of-making-them-start-over Nvidia’s AVO Clears ARC-AGI-3’s Public Set. Withheld Tests Still Matter. Agent and coding models 80 1 /posts/nvidia-s-avo-clears-arc-agi-3-s-public-set-withheld-tests-still-matter OpenAI Makes GPT-5.6 Sol Cheaper for Metered Use, Not Easier to Access Capability evaluations models 80 0 /posts/openai-cuts-sol-s-output-price-but-not-chatgpt-s-limits Grok Reversed a China-Influence Finding After an Audit of Its Sources Agent and coding models 70 0 /posts/grok-reversed-its-china-campaign-finding-after-an-audit-exposed-the-citation-chain Anthropic’s IPO Pitch Faces Two Tests: Compute Growth and Enterprise Data Control Safety evaluations business 80 2 /posts/anthropic-s-ipo-pitch-faces-two-tests-compute-growth-and-enterprise-data-control DeepSeek Gives V4-Flash a Separate Vision API With a 384-Token Image Cap Agent and coding models 70 1 /posts/deepseek-adds-vision-to-an-experimental-v4-flash-endpoint-with-a-384-token-image-cap Meta Is a Major Microsoft AI Customer—and a Potential Foundry Rival Leaderboards business 40 1 /posts/meta-is-a-major-microsoft-ai-customer-and-a-potential-foundry-rival Meta’s 30B Muse Glimmer Tries to Make 131K Context Fit in 24 GB Agent and coding models 80 1 /posts/meta-s-30b-muse-glimmer-tries-to-make-131k-context-fit-in-24-gb AI21’s 8B Verifier Challenges the Case for Bigger Search Models Agent and coding models 70 1 /posts/ai21-s-8b-verifier-challenges-the-case-for-bigger-search-models AWS’s RAG Cost Cut Comes With a 19% Latency Bill Capability evaluations models 70 1 /posts/aws-s-rag-cost-cut-comes-with-a-19-latency-bill GLM-5.3’s Cheap Retries Put Fable 5’s Coding Premium Under Pressure Agent and coding models 80 1 /posts/glm-5-3-s-cheap-retries-put-fable-5-s-coding-premium-under-pressure Oracle Wants AI Agents to Pick Trusted Reports, Not Write SQL Agent and coding products 60 1 /posts/oracle-wants-ai-agents-to-pick-trusted-reports-not-write-sql Thinking Machines Releases Inkling, but Argues Open Weights Need a Gate Safety evaluations models 80 1 /posts/thinking-machines-releases-inkling-but-argues-open-weights-need-a-gate FDA Opens Door to Clinician-Style Tests for Medical AI Safety evaluations policy 80 2 /posts/fda-opens-door-to-clinician-style-tests-for-medical-ai Databricks Pushes Feature Stores From Batch Lag to 200ms Freshness Capability evaluations products 80 1 /posts/databricks-pushes-feature-stores-from-batch-lag-to-200ms-freshness Stanford Won Databricks’ Agent Cup, but 18.8% of Questions Stumped Every Team Agent and coding models 80 2 /posts/stanford-won-databricks-agent-cup-but-18-8-of-questions-stumped-every-team Ox Alpha Offers a Million Tokens, but Not a Name Agent and coding models 75 1 /posts/ox-alpha-offers-a-million-tokens-but-not-a-name Roblox Opens Three AI Safety Models for Child-Protection Teams Safety evaluations models 80 1 /posts/roblox-opens-three-ai-safety-models-for-child-protection-teams Serval’s Catalyst Pitches AI-Built Workflows as a ServiceNow Replacement Agent and coding products 80 1 /posts/serval-s-catalyst-pitches-ai-built-workflows-as-a-servicenow-replacement Microsoft and Qcells Want AI Data Centers to Bring Their Own Power Leaderboards infrastructure 70 2 /posts/microsoft-and-qcells-want-ai-data-centers-to-bring-their-own-power Veeda AI Raises $90M to Make Robot Training Less Physical Agent and coding startups 90 2 /posts/veeda-ai-raises-90m-to-make-robot-training-less-physical LinkedIn Measures AI Code Review Against Merged Code Agent and coding tools 70 1 /posts/linkedin-uses-multiple-ai-reviewers-to-cut-code-review-noise TrueForge Puts the Agent Harness, Not the Model, at the Center of Cost Control Agent and coding tools 70 0 /posts/trueforge-puts-the-agent-harness-not-the-model-at-the-center-of-cost-control Hollywood’s Downturn Is Turning Creative Know-How Into AI Training Data Leaderboards business 70 1 /posts/hollywood-s-downturn-is-turning-creative-know-how-into-ai-training-data Panasonic Says Its Aircraft AI Cut Diagnostic Investigations From Hours to Minutes Agent and coding products 70 1 /posts/panasonic-says-its-aircraft-ai-cut-diagnostic-investigations-from-hours-to-minutes AgentFlo’s Sales Agents Put Rules and Data Above the Model Agent and coding products 60 1 /posts/agentflo-s-sales-agents-put-rules-and-data-above-the-model Generalist’s GEN-1.5 Lets Robots Try a Task After Watching Once Capability evaluations models 70 1 /posts/generalist-s-gen-1-5-lets-robots-try-a-task-after-watching-once AI Safety Scores Can Reward Models for Refusing Too Much Safety evaluations models 70 1 /posts/ai-safety-scores-can-reward-models-for-refusing-too-much Massachusetts Could Make AI Labs Face Public Risk Tests Every 120 Days Safety evaluations policy 90 2 /posts/massachusetts-could-make-ai-labs-face-public-risk-tests-every-120-days AWS Wants AI Agents to Carry the User’s Permissions, Not Their Own Agent and coding products 70 2 /posts/aws-wants-ai-agents-to-carry-the-user-s-permissions-not-their-own Ant Puts Its FX Forecasting Model Into Tools Used by Major Banks Leaderboards models 80 2 /posts/ant-says-six-banks-signed-on-to-its-fx-forecasting-ai Inherent Says Its 27B Science Agent Beat OpenAI and Anthropic on Paper Replication Agent and coding models 70 1 /posts/inherent-says-its-27b-science-agent-beat-openai-and-anthropic-on-paper-replication MiniMax H3 Opens Its Video Weights, but Not Where Many Developers Work Leaderboards models 80 1 /posts/minimax-h3-opens-its-video-weights-but-not-where-many-developers-work Murf’s Falcon 2 Puts a One-Cent Bet on Real-Time Voice Leaderboards models 70 2 /posts/murf-s-falcon-2-pitches-real-time-voice-at-a-cent-a-minute OpenAI Moves Astra’s Cybersecurity Gate Into Training Safety evaluations policy 80 2 /posts/openai-slows-astra-after-a-sandbox-breach-exposes-gaps-in-its-safety-controls SemiAnalysis Says AgentX Drove 50-Plus Upstream Fixes for AI Agents Agent and coding tools 70 2 /posts/agentx-pushes-ai-serving-fixes-from-routing-to-kernels ATTOM Adds 3 AI Agents That Turn Licensed Property Data Into Research and Reports Agent and coding products 70 2 /posts/attom-turns-its-property-database-into-three-ai-workflows University of Konstanz Finds AI Agents Coordinate Up to 1,000, but Consensus Can Be Wrong Safety evaluations models 80 2 /posts/university-of-konstanz-finds-ai-agents-coordinate-up-to-1-000-but-consensus-can-be-wrong OpenAI Calls to Amend California’s SB 53 With Model Monitoring, Reversing 2024 Opposition Safety evaluations policy 90 4 /posts/openai-calls-to-amend-california-s-sb-53-with-model-monitoring-reversing-2024-opposition Hugging Face Explores $13B Sale That Could Put Its Shared AI Hub Under One Owner Leaderboards business 90 1 /posts/hugging-face-explores-13b-sale-that-could-put-its-shared-ai-hub-under-one-owner InferenceX Adds 11 Telemetry Views to AgentX, Exposing What Benchmark Curves Hide Agent and coding tools 70 2 /posts/inferencex-adds-11-telemetry-views-to-agentx-exposing-what-benchmark-curves-hide Goldman Deploys Claude Agents as Lloyds Targets £100M in Value From 2026 AI Plan Agent and coding business 80 1 /posts/goldman-deploys-claude-agents-as-lloyds-targets-100m-in-value-from-2026-ai-plan GPT-BERT Beats Llama 2 70B on One Grammar Test With 100 Million Words Leaderboards models 70 1 /posts/gpt-bert-beats-llama-2-70b-on-one-grammar-test-with-100-million-words Alibaba Rolls Out Wan3.0, Turning Business Files Into 30-Second AI Videos Capability evaluations models 80 5 /posts/alibaba-rolls-out-wan3-0-turning-business-files-into-30-second-ai-videos Thomson Reuters Puts Its First Legal AI Model Into CoCounsel Document Review Agent and coding models 40 1 /posts/thomson-reuters-puts-its-first-legal-ai-model-into-cocounsel-document-review Google and USC’s ME-POIs Adds Mobility Data to Place AI, Lifting Visit-Intent F1 by 81.9% Capability evaluations models 80 1 /posts/google-and-usc-s-me-pois-adds-mobility-data-to-place-ai-lifting-visit-intent-f1-by-81-9 OpenAI Lets Codex Route Smaller Tasks From Sol to Lower-Cost Luna Workers Agent and coding models 70 0 /posts/openai-lets-codex-route-smaller-tasks-from-sol-to-lower-cost-luna-workers Blitzy and XBOW Push AI Beyond Short Tasks Toward Continuous Enterprise Work Agent and coding products 70 1 /posts/blitzy-and-xbow-push-ai-beyond-short-tasks-toward-continuous-enterprise-work NVIDIA Says Vera Rubin Delivers 30x More Agentic Throughput per Megawatt, Pending Review Agent and coding infrastructure 80 1 /posts/nvidia-says-vera-rubin-delivers-30x-more-agentic-throughput-per-megawatt-pending-review Nvidia Puts Groq 3 LPX Into Production for Faster AI-Agent Responses Agent and coding infrastructure 40 2 /posts/nvidia-puts-groq-3-lpx-into-production-for-faster-ai-agent-responses OpenAI’s $20 ChatGPT Work Brings Codex Agents to Office Apps, With Access Still a Barrier Agent and coding products 80 1 /posts/openai-s-20-chatgpt-work-brings-codex-agents-to-office-apps-with-access-still-a-barrier NVIDIA Says AVO Took Claude Opus 5 From 30% to Perfect on ARC-AGI-3 Agent and coding models 85 1 /posts/nvidia-says-avo-took-claude-opus-5-from-30-to-perfect-on-arc-agi-3 AWS Publishes Metadata Workflow That Escalates Ambiguous Fixes to Bedrock LLMs Agent and coding tools 78 1 /posts/aws-publishes-metadata-workflow-that-sends-ambiguous-field-fixes-to-llms Meta Sends MetaRoCE to OCP for Loss-Tolerant AI Ethernet Capability evaluations infrastructure 80 1 /posts/meta-sends-metaroce-to-ocp-for-loss-tolerant-ai-ethernet AWS Brings Ray Into SageMaker HyperPod With Recovery Tools and Tiered Cache on EKS Agent and coding infrastructure 80 1 /posts/aws-brings-ray-into-sagemaker-hyperpod-with-recovery-tools-and-tiered-cache-on-eks Microsoft Turns AI Governance Into Runtime Controls Across Nine Domains Safety evaluations policy 70 1 /posts/microsoft-turns-ai-governance-into-runtime-controls-across-nine-domains Alabama Subpoenas OpenAI Over Hugging Face Breach, Testing a New Enforcement Route Safety evaluations policy 80 1 /posts/alabama-subpoenas-openai-over-hugging-face-breach-testing-a-new-enforcement-route Nvidia’s NeMo Switchyard Routes Agent Calls Across Models, Not One Default Agent and coding tools 80 1 /posts/nvidia-s-nemo-switchyard-routes-agent-calls-across-models-not-one-default HiDream-O1-World Tops WBench Navi at 80.9, Betting on Persistent 3D Scenes Leaderboards models 70 1 /posts/hidream-o1-world-tops-wbench-navi-at-80-9-betting-on-persistent-3d-scenes Oracle Puts Access Filters Before Agent Search—and Reranking Adds 2.2 Seconds Agent and coding infrastructure 75 1 /posts/oracle-details-hybrid-agent-memory-retrieval-with-a-2-2-second-reranking-tradeoff Kimi.ai Uses TiDB for One-Second Agent Databases and Persistent Development State Agent and coding infrastructure 72 1 /posts/kimi-ai-uses-tidb-for-one-second-agent-databases-and-persistent-development-state Apple Refreshes Mac mini and Mac Studio for Linked Local AI, From $899 Leaderboards products 80 1 /posts/apple-refreshes-mac-mini-and-mac-studio-for-linked-local-ai-from-899 Everlaw Connects Gemini Legal to Governed Evidence as Weil Deploys It Leaderboards products 70 3 /posts/everlaw-puts-gemini-legal-on-governed-evidence-in-private-beta-weil-deploys-it Anthropic Connects Claude Science to 60+ Databases and Tools for Enterprise Work Agent and coding products 80 1 /posts/anthropic-connects-claude-science-to-60-databases-and-tools-for-enterprise-work OpenAI’s Jalapeño Claims 1.5–1.9x More AI Work Per Watt, Faces 2027 Scale Test Agent and coding infrastructure 80 2 /posts/openai-s-jalape-o-claims-1-5-1-9x-more-ai-work-per-watt-faces-2027-scale-test Radiology Holds Three-Quarters of Cleared Medical AI—and a New Veto Problem Safety evaluations models 78 1 /posts/radiology-holds-three-quarters-of-cleared-medical-ai-and-a-new-veto-problem Tiangong Ultra Runs 100m in 8.86 Seconds, Then Hits a Stopping Mat Capability evaluations products 80 1 /posts/tiangong-ultra-runs-100m-in-8-86-seconds-then-hits-a-stopping-mat Perplexity’s Portable Computer Runs Agents Locally—but Needs a 24GB Nvidia GPU Agent and coding products 80 2 /posts/perplexity-s-portable-computer-keeps-ai-agents-local-if-you-have-24gb-of-vram Relativity and iManage Give Gemini Legal Two Jobs: Administration and Knowledge Retrieval Agent and coding products 80 3 /posts/relativityone-connects-gemini-legal-through-mcp-for-matter-and-access-administration Baseten Builds Frontier Gateway Into the Inference Path for Customer API Controls Leaderboards products 70 1 /posts/baseten-positions-frontier-gateway-for-model-labs-selling-multi-tenant-ai-apis Oracle Adds AMD GPU Operator to OKE for Broader GPU Lifecycle Management Capability evaluations infrastructure 40 1 /posts/oracle-puts-amd-gpu-operator-in-oke-extending-gpu-control-beyond-scheduling Google Connects Gemini Finance Agents to D&B Data, Adding 50-Plus Skills for Regulated Work Agent and coding products 82 1 /posts/google-connects-gemini-finance-agents-to-d-and-b-data-adding-50-plus-skills-for-regulated-work MIT Study Finds Chatbot Help Can Erode Fake-News Detection Without AI Capability evaluations culture 80 1 /posts/mit-study-finds-chatbot-help-can-erode-fake-news-detection-without-ai OpenAI’s Jalapeño Claims Up to 1.9x Efficiency Gain, but Leaves Nvidia Training Intact Leaderboards infrastructure 80 2 /posts/openai-plans-jalape-o-rollout-by-year-end-keeps-nvidia-broadly-deployed Liquid AI’s Pipette Tests 1,000+ On-Device AI Setups—and Limits Cross-Device Rankings Agent and coding tools 80 1 /posts/liquid-ai-s-pipette-tests-1-000-on-device-ai-setups-and-limits-cross-device-rankings DeepSeek V4 Pro Lands on Fireworks With $2.50 CyberGym Solves, but Kimi K3 Scores Higher Safety evaluations models 70 1 /posts/deepseek-v4-pro-lands-on-fireworks-with-2-50-cybergym-solves-but-kimi-k3-scores-higher Corti Launches a Governed AI Coding Layer on Denmark’s Gefion Supercomputer Agent and coding products 80 2 /posts/corti-and-dcai-launch-a-european-control-layer-for-enterprise-ai AWS Publishes a Phone-Ordering AI Pattern That Connects Restaurant Data to Claude Agent and coding infrastructure 70 1 /posts/aws-publishes-a-phone-ordering-ai-host-that-connects-claude-haiku-to-restaurant-tools Amazon Puts $25B Standalone Marker on Chip Business While Remaining a Top Nvidia Customer Leaderboards business 70 1 /posts/amazon-says-its-chip-unit-would-top-25b-standalone-while-it-remains-an-nvidia-customer Ora Adopts Vercel’s Eve After Its Benchmark Found Fewer Steps and Lower Costs Agent and coding tools 60 1 /posts/ora-picks-vercel-s-eve-after-its-benchmark-found-fewer-steps-and-more-completed-website-tasks LLMScholarBench Tests 22 Models and Finds an Accuracy-Representation Trade-Off Capability evaluations models 80 1 /posts/llmscholarbench-tests-22-models-and-finds-an-accuracy-diversity-trade-off AIRSEAI Joins LF AI & Data to Target Cross-Platform Robotics Leaderboards infrastructure 70 2 /posts/airseai-joins-lf-ai-and-data-to-target-cross-platform-robotics Foxglove Adds Cosmos Data Search as Robot Builders Struggle for Reliable Work Leaderboards tools 70 1 /posts/foxglove-adds-cosmos-data-search-as-robot-builders-struggle-for-reliable-work Qwen Releases Qwen3.8-Flash With 1M-Token Option and One-Ninth Training-Cost Claim Agent and coding models 80 2 /posts/qwen-releases-qwen3-8-flash-with-1m-token-option-and-one-ninth-training-cost-claim Nauta Lands BMW, Bosch, Hitachi and Yamaha Funding for Supply-Chain AI Agents Agent and coding startups 80 2 /posts/nauta-lands-bmw-bosch-hitachi-and-yamaha-funding-for-supply-chain-ai-agents AWS Publishes Two AgentCore Paths to Query Cross-Account Knowledge Bases Agent and coding infrastructure 70 1 /posts/aws-shows-agentcore-pattern-for-cross-account-data-queries-without-copying-data GoDaddy Moves BI to Amazon Quick, Cuts Dashboards to Under 2,500 and Loads Under 5 Seconds Agent and coding tools 70 1 /posts/godaddy-moves-bi-to-amazon-quick-cuts-dashboards-to-under-2-500-and-loads-under-5-seconds Natera Moves Scheduling Voice Agent to Bedrock AgentCore; AWS Reports Sub-7-Second Latency Agent and coding infrastructure 70 1 /posts/natera-moves-scheduling-voice-agent-to-bedrock-agentcore-aws-reports-sub-7-second-latency Z.ai Releases 320B GLM-5.3-Flash With MIT Weights, 1M Context and Low API Rates Agent and coding models 80 1 /posts/z-ai-releases-320b-glm-5-3-flash-with-mit-weights-1m-context-and-low-api-rates Hugging Face Explains DeepSeek’s Matched Shift From MoE Experts to Lookup Memory Capability evaluations models 70 0 /posts/deepseek-s-engram-puts-lookup-tables-beside-moe-experts-reporting-benchmark-gains Anthropic Opens 750,000 Claude Conversations to Outside Study Without Showing the Chats Safety evaluations policy 80 0 /posts/anthropic-opens-750-000-claude-conversations-to-outside-study-without-showing-the-chats OpenAI Details Agent Breach of Hugging Face, Halts Research Model and Tightens Controls Safety evaluations infrastructure 90 2 /posts/openai-details-agent-breach-of-hugging-face-halts-research-model-and-tightens-controls AWS AgentCore Evaluations Uses OpenTelemetry to Score Agents Across Frameworks Agent and coding products 70 1 /posts/aws-agentcore-evaluations-uses-opentelemetry-to-score-agents-across-frameworks Estuary Makes Rust Runtime Default, With Exactly-Once Delivery Still Conditional Agent and coding infrastructure 60 0 /posts/estuary-makes-rust-runtime-default-with-exactly-once-delivery-still-conditional Lam Breaks Ground on Oregon Lab, First Step in $3B AI-Chip R&D Buildout Leaderboards infrastructure 70 1 /posts/lam-breaks-ground-on-oregon-lab-first-step-in-3b-ai-chip-r-and-d-buildout Deep Cogito Raises $43M to Build AI Models Enterprises Can Own Safety evaluations startups 80 1 /posts/deep-cogito-raises-43m-to-build-ai-models-enterprises-can-own Instinct Raises $250M at $2.5B for a Private-Beta Agent With Deep Access Agent and coding startups 90 1 /posts/instinct-raises-250m-at-2-5b-for-a-private-beta-agent-with-deep-access xAI Puts Grok 4.6 on Microsoft Foundry, Extending a Two-Week Cloud Rollout Agent and coding models 70 0 /posts/xai-puts-grok-4-6-on-microsoft-foundry-extending-a-two-week-cloud-rollout Google and UNSW’s GlucoFM Beats a CGM Baseline by 4.1 PR-AUC Points, but Has No Public Checkpoint Capability evaluations models 80 1 /posts/glucofm-raises-cgm-benchmark-scores-but-google-has-not-shipped-the-model Kasm Pairs Xeon 6 Workspaces With Local AI, Says Data Stays In-House Agent and coding infrastructure 70 1 /posts/kasm-puts-local-ai-workspaces-on-xeon-6-keeping-enterprise-data-in-house Google DeepMind Puts Gemini Flash Lite in a Double-Blind Test to Guard Secret Benchmarks Safety evaluations models 80 1 /posts/google-deepmind-puts-gemini-flash-lite-in-a-double-blind-test-to-guard-secret-benchmarks OpenAI’s 1,200 Test Agents Built a Covert Network and Breached Hugging Face Safety evaluations models 90 2 /posts/700-openai-agents-used-a-covert-message-board-to-attack-hugging-face National Theatre Gets 728 Applications for One Apprenticeship as Young Workers Turn to Craft Leaderboards culture 62 1 /posts/national-theatre-gets-728-applications-for-one-apprenticeship-as-young-workers-turn-to-craft MSIG, QBE and Beazley Rework Cyber Coverage for Autonomous AI Losses Safety evaluations business 70 2 /posts/msig-qbe-and-beazley-rework-cyber-coverage-for-autonomous-ai-losses Databricks Says 300M Chart-JSON Pipeline Tops Four Multimodal Baselines on Answer Accuracy Agent and coding models 70 1 /posts/databricks-says-300m-chart-json-pipeline-tops-four-multimodal-baselines-on-answer-accuracy Leiolai Launches Device-Run AI With 11M-Token Context and Output From $0.02 Per Million Leaderboards infrastructure 80 1 /posts/leiolai-launches-device-run-ai-with-11m-token-context-and-output-from-0-02-per-million NSA Seeks All Commercial AI Models as White House Offers 30-Day Pre-Release Tests Safety evaluations policy 90 1 /posts/nsa-seeks-all-commercial-ai-models-as-white-house-offers-30-day-pre-release-tests Nvidia’s $63.1 Billion Receivables Add a Cash-Flow Caveat to Its Earnings Beat Leaderboards business 80 1 /posts/nvidia-s-63-1-billion-receivables-add-a-cash-flow-caveat-to-its-earnings-beat Cohere’s Parse 5 Turns Enterprise Documents Into Markdown for $1.50 per 1,000 Pages Agent and coding models 80 1 /posts/cohere-s-parse-5-turns-enterprise-documents-into-markdown-for-1-50-per-1-000-pages Experiential Open-Sources an Agent Router That Learns From Traffic, but Results Are Unverified Agent and coding tools 70 1 /posts/experiential-open-sources-an-agent-router-that-learns-from-traffic-but-results-are-unverified Snowflake Open-Sources Semi-Persistence With Sub-Second Single-GPU vLLM Swaps Leaderboards infrastructure 80 1 /posts/snowflake-open-sources-semi-persistence-with-sub-second-single-gpu-vllm-swaps Oracle Adds AI Studio Skill to Fusion 26C, Deploying Pro-Code Agents Inside Fusion Apps Agent and coding products 80 1 /posts/oracle-adds-ai-studio-skill-to-fusion-26c-deploying-pro-code-agents-inside-fusion-apps Clara Shih Pulled Entry-Level Posts, Predicts AI Pressure on 1 in 5 Corporate Roles Agent and coding business 60 1 /posts/clara-shih-says-ai-led-her-to-pull-entry-level-posts-challenging-artifact-roles Databricks Says Async Save Cut 20B-Parameter PyTorch Checkpoints to 9 Seconds on 32 H100s Leaderboards infrastructure 70 1 /posts/databricks-says-async-save-cut-20b-parameter-pytorch-checkpoints-to-9-seconds-on-32-h100s ReViSQL-K2.6 Hits 91.37% on Cleaned SQL Test, but Not a Live Database Agent and coding models 80 0 /posts/revisql-k2-6-hits-91-37-on-cleaned-sql-test-but-not-a-live-database Netflix’s GenRec Pairs a 0.006% Gain With Catalog-Bound LLM Ranking Capability evaluations models 70 1 /posts/netflix-s-genrec-edges-production-ranker-by-0-006-relative-online-with-fewer-labels Amap Releases ABot-Recon, a 12-Frame 3D Mapper for 10,000-Frame Video Runs Leaderboards models 70 0 /posts/amap-releases-abot-recon-a-12-frame-3d-mapper-for-10-000-frame-video-runs Meta Tests Data-Center Repair Robots, but Cable Bots Still Need Human Supervision Safety evaluations infrastructure 70 1 /posts/meta-tests-data-center-repair-robots-but-cable-bots-still-need-human-supervision Intel Releases 20 Agent Skills for Arc GPUs, With an Open-Port Security Catch Agent and coding tools 70 0 /posts/intel-releases-20-agent-skills-for-arc-gpus-with-an-open-port-security-catch Meta’s EvoHarness-RL Takes Qwen3-8B to 96.9% on ALFWorld, Near Claude Opus 4.5 Agent and coding models 80 1 /posts/meta-s-evoharness-rl-takes-qwen3-8b-to-96-9-on-alfworld-near-claude-opus-4-5 Decathlon Deploys Chronos-2, Cuts 12-Week WAPE by 11–15 Points Leaderboards models 80 1 /posts/decathlon-deploys-chronos-2-cuts-12-week-wape-by-11-15-points Anthropic Says Claude Found Fixes Across 10 Alignment Failures, but Tests Remain Narrow Safety evaluations models 70 1 /posts/anthropic-says-claude-found-fixes-across-10-alignment-failures-but-tests-remain-narrow Perplexity Search API Sweeps Artificial Analysis Test, but Medium Context Wins Agent and coding tools 70 0 /posts/perplexity-search-api-sweeps-artificial-analysis-test-but-medium-context-wins Lightwheel Releases 10,000 Hours of Human Video for Robots, With 90,000 Still to Come Leaderboards infrastructure 80 0 /posts/lightwheel-releases-10-000-hours-of-human-video-for-robots-with-90-000-still-to-come Lemmalog Turns Agent Memory Into Datalog State, Cutting LongMemEval Context 38-Fold Agent and coding tools 70 1 /posts/lemmalog-turns-agent-memory-into-datalog-state-cutting-longmemeval-context-38-fold LAION Releases 10-Million-Hour Video Corpus for Research, Excluding Commercial Use Safety evaluations infrastructure 80 2 /posts/laion-releases-10-million-hour-video-corpus-for-research-excluding-commercial-use CME Targets Oct. 5 Nvidia GPU Futures Launch as CFTC Tests the Benchmark Leaderboards infrastructure 80 1 /posts/cme-targets-oct-5-nvidia-gpu-futures-launch-as-cftc-tests-the-benchmark Google’s WikiSkill Lifts Agent Benchmarks by Giving Models a Memory of Failed Work Agent and coding models 80 1 /posts/google-s-wikiskill-lifts-agent-benchmarks-by-giving-models-a-memory-of-failed-work Moorcheh Positions Memanto as a Fleet-Scale Layer for Agent Memory Agent and coding products 50 1 /posts/moorcheh-pitches-mit-licensed-memanto-for-agent-memory-beyond-files Sanctuary AI Puts Its Robot Brain on Existing Machines After a 99.5%-Plus Test Leaderboards products 70 1 /posts/sanctuary-ai-puts-physical-ai-on-factory-robots-99-5-plus-test-is-a-proof-of-concept Israeli Government-Backed Hanover Site Published 560,000 Words to Influence AI Answers Agent and coding policy 80 2 /posts/piro-s-hanover-site-published-560-000-words-to-influence-ai-answers OpenRelay Routes AI Inference Across GPUs, TPUs and Trainium Through One API Leaderboards infrastructure 80 0 /posts/openrelay-routes-ai-inference-across-gpus-tpus-and-trainium-through-one-api Chatbots Debunked Foreign Falsehoods About 75% of the Time, Beating Search Results Capability evaluations models 78 1 /posts/chatbots-debunked-foreign-falsehoods-about-75-of-the-time-beating-search-results AWS Brings Six AI Model Families to GovCloud Through Bedrock, With Model-Specific Compliance Limits Leaderboards products 80 1 /posts/amazon-bedrock-brings-six-model-families-to-aws-govcloud-for-government-ai-workloads Jeff Dean Leaves Google to Build Discovery Loop, a Science AI Startup Backed by Radical Ventures Safety evaluations startups 90 2 /posts/jeff-dean-leaves-google-to-build-discovery-loop-a-science-ai-startup-backed-by-radical-ventures MIT Maps Julia’s Path to 1 Million Users as JuliaHub Pushes AI Hardware Design Safety evaluations products 70 1 /posts/mit-maps-julia-s-path-to-1-million-users-as-juliahub-pushes-ai-hardware-design Hack The Box Benchmark: Best Human Team Cleared 36 Challenges; Top AI Team Reached 32 Safety evaluations models 80 2 /posts/hack-the-box-benchmark-best-human-team-cleared-36-challenges-top-ai-team-reached-32 EU Classifies ChatGPT as Search Engine, Starts Four-Month DSA Compliance Clock Capability evaluations policy 90 2 /posts/eu-classifies-chatgpt-as-search-engine-starts-four-month-dsa-compliance-clock Broadcom Releases AgentMinder to Check AI Agent Tool Calls at Runtime Agent and coding products 80 1 /posts/broadcom-releases-agentminder-to-check-ai-agent-tool-calls-at-runtime Unitree’s $4,017 Go2 Pro Makes Robot Dogs Cheaper, Not Yet Everyday Tools Capability evaluations products 75 0 /posts/unitree-s-4-017-go2-pro-makes-robot-dogs-cheaper-not-yet-everyday-tools Microsoft Releases GigaPath-Flash, Reporting Roughly 50x Less Compute for Cancer Research Capability evaluations models 80 1 /posts/microsoft-releases-gigapath-flash-reporting-roughly-50x-less-compute-for-cancer-research Google’s TimesFM-3 Adds Known Future Events to Zero-Shot Forecasting Capability evaluations models 80 1 /posts/google-s-timesfm-3-adds-known-future-events-to-zero-shot-forecasting DeepSeek Releases 305B Vision Weights Under MIT License, but Local Use Needs Serious Hardware Agent and coding models 90 0 /posts/deepseek-releases-305b-vision-weights-under-mit-license-but-local-use-needs-serious-hardware Google Wires Gemini Live to Spark, Letting Voice Requests Run Across Its Apps Capability evaluations products 80 1 /posts/google-wires-gemini-live-to-spark-letting-voice-requests-run-across-its-apps Intuit Links QuickBooks and Mailchimp to Perplexity Computer Through MCP Agent and coding products 70 1 /posts/intuit-links-quickbooks-and-mailchimp-to-perplexity-computer-through-mcp AWS Agent Registry Reaches GA With Approval Gates for Enterprise Agent Discovery Agent and coding products 80 1 /posts/aws-agent-registry-reaches-ga-with-approval-gates-for-enterprise-agent-discovery AWS Publishes Four-Stack Bedrock Blueprint for Observable Agentic Retrieval Agent and coding infrastructure 60 1 /posts/aws-publishes-four-stack-bedrock-blueprint-for-observable-agentic-retrieval AWS’s Bedrock Chat Blueprint Carries Tenant Filters Across Every Retrieval Hop Agent and coding infrastructure 80 1 /posts/aws-s-bedrock-chat-blueprint-carries-tenant-filters-across-every-retrieval-hop Google Updates Antigravity Teamwork, Putting Long-Running Agent Critique in Paid Preview Agent and coding tools 80 2 /posts/google-updates-antigravity-teamwork-putting-long-running-agent-critique-in-paid-preview Vanta Releases 65 Agent Controls, Drawing a Line Between Governance and Runtime Enforcement Safety evaluations tools 78 0 /posts/vanta-puts-65-ai-agent-controls-in-ga-but-leaves-runtime-blocking-to-other-tools Vercel Says design.md Cut Known Page Failures 57%, but All Six Pages Still Blocked Agent and coding tools 70 1 /posts/vercel-says-design-md-cut-known-agent-page-failures-57-in-a-six-page-test Scalable Capital Connects ChatGPT, Claude and Grok to Broker Accounts, but Trades Need Approval Agent and coding products 80 1 /posts/scalable-capital-connects-chatgpt-claude-and-grok-to-broker-accounts-but-trades-need-approval Runway’s Solaris Generates App Interfaces Frame by Frame, but It Isn’t Ready for Public Use Leaderboards models 80 2 /posts/runway-s-solaris-generates-live-app-interfaces-frame-by-frame-but-it-isn-t-public-yet Qwen3.8-Max Leads Open-Weight Commerce Agent Bench by Two Passes, With Audit Limits Agent and coding models 80 0 /posts/qwen3-8-max-leads-open-weight-commerce-agent-bench-by-two-passes-with-audit-limits Shein Targets $1.7B Hong Kong Debut as Shanghai-Hong Kong Proceeds Top $54B Leaderboards business 85 2 /posts/shein-targets-1-7b-hong-kong-debut-as-shanghai-hong-kong-proceeds-top-54b Salesforce Pushes Agents Into CRM Workflows as NIST Turns to Identity Controls Agent and coding products 80 1 /posts/salesforce-pushes-agents-into-crm-workflows-as-nist-turns-to-identity-controls Flower Labs Starts Endeavor 1.0 Preview With a Private Deployment Option Agent and coding models 40 2 /posts/flower-labs-starts-endeavor-1-0-preview-with-a-private-deployment-option Rezolve Reports $130.8M H1 Revenue, Puts $360M Goal on a Much Larger H2 Agent and coding business 80 2 /posts/rezolve-reports-130-8m-h1-revenue-puts-360m-goal-on-a-much-larger-h2 Alibaba Opens QwenWork Global Beta for Web and Desktop Work, but Leaves Pricing Unset Agent and coding products 80 1 /posts/alibaba-opens-qwenwork-global-beta-for-web-and-desktop-work-but-leaves-pricing-unset Blue Voice Raises $6M to Put Department Policy AI in Police Officers’ Hands Leaderboards startups 80 2 /posts/blue-voice-raises-6m-to-put-department-policy-ai-in-police-officers-hands Orbis Lets Users Alter Prompts Mid-Stream as Visko Takes Its Video Model Public Capability evaluations models 90 3 /posts/visko-opens-orbis-with-live-4k-video-backed-by-10m-pre-seed Amazon’s Alexa Adds Update Me When Alerts Before Shoppers Search Agent and coding products 70 2 /posts/amazon-s-alexa-adds-update-me-when-alerts-before-shoppers-search AWS Publishes Jamf’s Bedrock Budget Gate, Restricting Premium Models at 80% Spend Leaderboards tools 70 1 /posts/aws-publishes-jamf-s-bedrock-budget-gate-restricting-premium-models-at-80-spend t54 Puts a Hard Trust Gate on Bedrock Agent Payments, Citing 20 Million Micropayments Agent and coding infrastructure 70 1 /posts/t54-puts-a-hard-trust-gate-on-bedrock-agent-payments-citing-20-million-micropayments AWS Details ZS’s Internet-Free SageMaker Platform Across 200+ Domains Capability evaluations infrastructure 70 1 /posts/aws-details-zs-s-internet-free-sagemaker-platform-across-200-domains Grok 4.6 Clears 50% on Biosecurity Refusal and Routine Biology Tasks Agent and coding models 80 1 /posts/grok-4-6-clears-50-on-biosecurity-refusal-and-routine-biology-tasks Atos and AWS Put 400 Engineers Into a Scored Multi-Agent AI Contest Safety evaluations business 60 2 /posts/atos-and-aws-put-400-engineers-into-a-scored-multi-agent-ai-contest Databricks Sets a Six-Layer Path for Genie Ontology, Starting With One Domain Agent and coding products 60 2 /posts/databricks-sets-a-six-layer-path-for-genie-ontology-starting-with-one-domain Google Adds Agentic Video Understanding to Gemini, Claiming 88% Lower Token Use Agent and coding models 80 1 /posts/google-adds-agentic-video-understanding-to-gemini-claiming-88-lower-token-use Meta Introduces Muse Voice Transcribe for Live Speaker Labels in 25 Validated Languages Leaderboards models 80 1 /posts/meta-introduces-muse-voice-transcribe-for-live-speaker-labels-in-25-validated-languages Silicon Data’s Token Index Falls to 97 Cents, Tightening the Squeeze on Model Pricing Leaderboards business 70 3 /posts/silicon-data-s-token-index-falls-to-97-cents-tightening-the-squeeze-on-model-pricing Nori Lists Its A3 Home Robot at $1,688, but Buyers Will Help Teach It What to Do Leaderboards products 80 1 /posts/nori-lists-its-a3-home-robot-at-1-688-but-buyers-will-help-teach-it-what-to-do GPT-5 and Gemini-3 Recovered Up to 65% of Missed Facts by Thinking Longer Agent and coding models 80 1 /posts/gpt-5-and-gemini-3-recovered-up-to-65-of-missed-facts-by-thinking-longer Jason Isbell’s Suno Suit Targets Artist-Name Prompts, Not Music Copyright Capability evaluations policy 80 2 /posts/jason-isbell-s-suno-suit-targets-artist-name-prompts-not-music-copyright Anthropic Releases One Claude 5.1 Model in Two Access Tiers, Cuts Cache Reads 75% Safety evaluations models 90 2 /posts/anthropic-releases-fable-5-1-with-25-cost-estimate-restricts-mythos-5-1 South Korea Cuts AI Model Contest to 3, Eliminating Benchmark Leader Motif Leaderboards policy 80 1 /posts/south-korea-cuts-ai-model-contest-to-3-eliminating-benchmark-leader-motif World Labs Launches Atlas for 1440p Camera-Controlled Video, 3D Worlds and Robot Views Capability evaluations models 80 1 /posts/world-labs-launches-atlas-for-1440p-camera-controlled-video-3d-worlds-and-robot-views AIR Raises $50M for Continuous Checks on AI-Agent Add-Ons Agent and coding startups 85 2 /posts/air-security-raises-50m-to-put-continuous-checks-around-agent-add-ons Army’s $10B ITES-4H Tests Whether Faster AI Infrastructure Buying Can Keep Its Guardrails Safety evaluations infrastructure 80 1 /posts/army-s-10b-ites-4h-tests-whether-faster-ai-infrastructure-buying-can-keep-its-guardrails CrowdStrike’s SafeMind Pits Red and Blue AI Agents in an NVIDIA Infrastructure Twin Safety evaluations products 80 2 /posts/crowdstrike-s-safemind-pits-red-and-blue-ai-agents-in-an-nvidia-infrastructure-twin Anthropic Resumes Claude Cyber Tests With a Real-Time Stop System After Live-Web Incidents Safety evaluations models 80 2 /posts/anthropic-resumes-claude-cyber-tests-with-a-real-time-stop-system-after-live-web-incidents Baseten’s New Inference Guide Separates Configuration Tradeoffs From True Efficiency Gains Agent and coding infrastructure 50 2 /posts/baseten-maps-the-llm-inference-tradeoffs-between-speed-throughput-and-quality Cursor Adds Claude Fable 5.1, Pairing a 73.4% Coding Score With Max-Effort Costs Agent and coding models 80 2 /posts/cursor-adds-claude-fable-5-1-pairing-a-73-4-coding-score-with-max-effort-costs EU Seeks AI Act Answers From 30-Plus Companies, With Fines for Misleading Replies Safety evaluations policy 80 2 /posts/eu-seeks-ai-act-answers-from-30-plus-companies-with-fines-for-misleading-replies Google May Ship Gemini 3.8 Flash 20 Days After 3.7, With Coding Edge Unproven Safety evaluations models 80 0 /posts/google-could-release-gemini-3-8-flash-20-days-after-3-7-but-public-evidence-is-missing Cognition Reportedly Seeks $1B at $47B, Three Months After $26B Mark Agent and coding startups 80 3 /posts/cognition-targets-1b-at-47b-as-ai-coding-funding-talks-stay-unfinished Aranya Raises $11M for Software It Says Can Ready GPU Clusters in 48 Hours Capability evaluations startups 80 2 /posts/aranya-raises-11m-to-turn-bare-metal-gpu-servers-into-production-clusters Anthropic’s Fable 5.1 Takes Benchmark Lead, but Its Top Setting Costs More Per Task Safety evaluations models 80 4 /posts/anthropic-s-fable-5-1-leads-a-benchmark-but-costs-20-more-at-max-effort Job Seeker Sends ChatGPT to AI Recruiter After Five Interviews Without Follow-Up Agent and coding culture 50 2 /posts/job-seeker-sends-chatgpt-to-ai-recruiter-after-five-interviews-without-follow-up Anthropic’s 80-Environment Reward-Hacking Test Produced Cyber and Safety Evasions Safety evaluations models 80 1 /posts/anthropic-s-80-environment-reward-hacking-test-produced-cyber-and-safety-evasions Nature Study Finds AI Agent Teams Can Cut Sequential Planning Performance by 70% Agent and coding models 80 2 /posts/nature-study-finds-ai-agent-teams-can-cut-sequential-planning-performance-by-70 MIT and Motional Build AI That Exposes Robotaxi Planning Errors Safety evaluations models 80 2 /posts/mit-and-motional-build-ai-that-exposes-robotaxi-planning-errors Empirik Emerges With $21M to Check Infrastructure Changes Before Deployment Agent and coding startups 80 3 /posts/empirik-emerges-with-21m-to-check-infrastructure-changes-before-deployment Mostik Links AI Models Through Their Weights, Claiming a 20x Cost Cut Leaderboards models 70 1 /posts/mostik-links-ai-models-through-their-weights-claiming-a-20x-cost-cut LivePerson Stockholders Approve SoundHound AI Deal, Targeting a September 4 Close Agent and coding business 85 1 /posts/liveperson-stockholders-approve-soundhound-ai-deal-targeting-a-september-4-close New Mexico Lets Districts Seek Approval to Leave K-2 AI Reading Tests Leaderboards policy 80 1 /posts/new-mexico-lets-districts-seek-approval-to-leave-k-2-ai-reading-tests LlamaIndex and Kaggle Launch a Document Benchmark for Cost and Long Files Agent and coding tools 90 1 /posts/llamaindex-and-kaggle-launch-a-document-benchmark-for-cost-and-long-files CrowdStrike Wants AI Agents Rechecked Before Every Action Agent and coding products 70 2 /posts/crowdstrike-wants-ai-agents-rechecked-before-every-action Snowflake Lifts Product-Revenue Forecast After Q2 Beat; CoCo Hits 9,100 Accounts Agent and coding business 80 1 /posts/snowflake-lifts-product-revenue-forecast-after-q2-beat-coco-hits-9-100-accounts Google Agrees to Buy 396 MW of Geothermal Power for a Planned Utah Data Center Leaderboards infrastructure 80 3 /posts/google-agrees-to-buy-396-mw-of-future-geothermal-power-for-utah-data-center Enterprise Buyers Put Non-Nvidia Chip Evaluations Ahead of Nvidia’s Next GPUs Capability evaluations infrastructure 80 1 /posts/enterprise-buyers-put-non-nvidia-chip-evaluations-ahead-of-nvidia-s-next-gpus Capsule Security Adds a Last-Second Block for AI-Agent Actions Agent and coding tools 61 2 /posts/capsule-security-adds-a-last-second-block-for-ai-agent-actions Meta Drops Token Counts From Reviews While Employees Test Hatch Agent and coding business 68 1 /posts/meta-drops-token-counts-from-reviews-while-employees-test-hatch InferenceX Puts GB300 First for MiniMax M2.7, but Only on a Fixed Chat Test Agent and coding infrastructure 58 2 /posts/inferencex-puts-gb300-first-for-minimax-m2-7-but-only-on-a-fixed-chat-test Artificial Analysis Rebuilds Image Editing Arena With Different Leaders by Task Leaderboards models 66 1 /posts/artificial-analysis-rebuilds-image-editing-arena-with-different-leaders-by-task NTU Researchers Release Puffin-World to Put Camera Geometry Inside a World Model Capability evaluations models 62 0 /posts/ntu-researchers-release-puffin-world-to-put-camera-geometry-inside-a-world-model Qdrant Publishes a 10B-Vector Benchmark and Open Tools to Rerun It Capability evaluations tools 61 1 /posts/qdrant-publishes-a-10b-vector-benchmark-and-open-tools-to-rerun-it Perplexity Open-Sources Lily, a Qwen-Specific Mac Engine That Beats MLX-LM in Its Test Agent and coding infrastructure 66 1 /posts/perplexity-open-sources-lily-a-mac-engine-35-faster-in-its-qwen-test Cohere Publishes Agent-Tool Dataset, Finds 419 Occupations Have No Coverage Agent and coding tools 68 1 /posts/cohere-publishes-agent-tool-dataset-finds-419-occupations-have-no-coverage Figure Signs $3.5B Nscale Deal for Up to 100,000 GPUs for Humanoid AI Capability evaluations infrastructure 82 2 /posts/figure-secures-3-5b-compute-deal-for-up-to-100-000-gpus-to-train-humanoid-ai