Archive
Browse past daily topics
Every published daily topic, grouped by category. Use these to catch up on angles you missed or revisit evergreen framing.
Product releases
54 topics- Google retools CC as an agent for whole families — its own Google account, up to six members, US 18+ by invite or waitlistSep 19, 20262 sources
- Instinct and Meta's Muse add phone calling the same day — a Concierge early access here, a request-gated beta thereSep 18, 20263 sources
- Pinterest puts AI room redesigns one photo away — Restyle lands as an early preview in the US and Canada, broader next monthSep 18, 20262 sources
- Anthropic merges Claude chat and Cowork into one — Claude Docs and Claude Slides land in beta on paid plans, rolling out to Pro and Max firstSep 17, 20263 sources
- Threads builds a podcast promo kit — banners, episode cards, transcripts and guest cards, with no hosting in the pictureSep 17, 20262 sources
- Meta One turns creator tools into a subscription ladder — from $2.99 single apps up to a Max tier starting at $499/month, with links in reels gated behind the $49.99-and-up planSep 16, 20262 sources
- OpenAI pauses new $200 ChatGPT Pro subscriptions under Astra demand — existing accounts untouched, no end date givenSep 12, 20263 sources
- Google ships a Gemini app for Windows — Alt+Space drops it over your work, with Spark, Nano Banana, and Omni insideSep 11, 20262 sources
- Apple Watch will replay what you just heard: Audio Intelligence's official claims, privacy boundaries, and beta timelineSep 10, 20263 sources
- iPhone 18 Pro will attach a 'digital negative' to your photos: what Reference Image proves — and what it doesn'tSep 10, 20263 sources
- Suno v6 is a clean break: licensed-music training, legacy models headed for retirement — the migration window is openSep 10, 20262 sources
- Anthropic warns that infostealers are draining Claude subscribers' tokens — and every receipt is publicSep 9, 20263 sources
- Meta launches Muse — a personal agent that opens browsers, fills out forms, and checks out with Stripe, wrapped in security claims that haven't been tested yetSep 9, 20264 sources
- Instagram renames its AI label 'AI-generated profile' — and unlabeled AI profiles may lose reachSep 1, 20262 sources
- Google Gemini 3.7 Flash: coding + agent workhorse lands 3 weeks after 3.6 Flash, with two-tier pricingAug 14, 20263 sources
- OpenAI + Cerebras: GPT-5.6 Sol Ultrafast mode on the OpenAI API — 750 t/s, 11x faster than Fable 5Aug 14, 20262 sources
- Mistral OCR 4.1: the Document AI surface that gets competitive attention in the 8/13-8/14 frontier pricing cycleAug 14, 20262 sources
- xAI Grok 4.6: next major frontier model lands with Artificial Analysis Intelligence Index 61Aug 13, 20262 sources
- OpenAI Codex desktop for Linux enters preview — the macOS-CLI-only era endsAug 13, 20262 sources
- Google Flash trio + Robotics ER 2 + Lyria 3.5 surface in the July 2026 recap — what creators should pick up from Google's developer-focused monthAug 12, 20262 sources
- Meta Muse Glimmer: 30B Apache 2.0 open-weight model lands for on-device agentic workflowsAug 12, 20263 sources
- World models for creators: 3D scene asset vs interactive experience — when to use RTFM vs when to use Genie 3Aug 11, 20263 sources
- Character consistency over shots: Runway Gen-4.5 + Sora 2 — the property that makes serial video content shippableAug 11, 20263 sources
- Permission-aware enterprise search: Glean + Notion — the property that makes cross-tool search respect security boundariesAug 11, 20263 sources
- PostgreSQL 17 GA + Neon branching GA: self-hosted PG 17 vs serverless Postgres — the choice that defines 2026 database strategyAug 11, 20263 sources
- Real-time 3D + interactive world models: RTFM + Genie 3 — the property that makes world models creator-researchableAug 11, 20263 sources
- Video control surfaces: Runway Act-Two vs Pika Pikaframes — motion-capture vs keyframe-directorialAug 11, 20263 sources
- Supabase + Cloudflare D1: edge-runtime databases — full-stack Postgres vs global SQLiteAug 11, 20263 sources
- Native audio video generation: Veo 3.5 + Sora 2 — the property that removes the separate-audio-pipeline stepAug 11, 20263 sources
- Video generation cluster: Runway Gen-4.5 + Veo 3.5 + Sora 2 + Pika 2.2 — every video model ships a longer-coherent / native-audio milestone in one weekAug 11, 20266 sources
- VLA / robotics foundation models: NVIDIA Cosmos + Figure Helix 2 — general-purpose robotics vs humanoid-specific VLAAug 11, 20263 sources
- AI CLI tools 1.0: Claude Code + Codex CLI + Gemini CLI — three terminal-first agent surfaces, three different betsAug 7, 20264 sources
- CodeRabbit + Graphite: AI code review — rich inline review vs layered reviewers that can disagreeAug 7, 20263 sources
- OpenAI Agents SDK 1.0 + Anthropic Agent SDK 1.0: the vendor agent SDK pair — handoffs vs Skills, guardrails vs computer useAug 6, 20263 sources
- Agent observability: LangGraph Studio + OpenAI Traces + Anthropic Workbench — three debugging surfaces ship in the same weekAug 6, 20264 sources
- Anthropic 7/14 Claude for Teachers: free Claude for US K-12 educators + Learning Commons + 10 ed-tech platform integrationsAug 5, 20262 sources
- Claude 4.5 Sonnet GA: programmatic tool calling + web_fetch + 1M-token context hit general availabilityAug 5, 20262 sources
- Figma Make + Notion AI Agents + Linear AI: the 'productivity-tool agent surface' pattern, three examplesAug 5, 20264 sources
- Llama Guard 4 + C2PA Content Credentials 2.0: the creator-facing safety stack — input/output filtering + media provenance, both open-weight / open-standardAug 5, 20263 sources
- Groq LPU v3: deterministic token-stream latency — the property real-time voice and agent loops actually needAug 5, 20262 sources
- Qwen3-Omni + DeepSeek V3.2: the multimodal + long-context open-weight pairAug 5, 20263 sources
- Apple MLX 1.0 + MLX-LM: on-device LLM inference on Apple Silicon — privacy-preserving creators get a real frameworkAug 5, 20262 sources
- OpenAI Realtime GA + Realtime-mini tier: production-grade voice agents cross the always-on thresholdAug 5, 20262 sources
- Scheduled Tasks + Claude Skills + Vercel AI Gateway: the always-on agent runtime stack that replaces bespoke infraAug 5, 20264 sources
- HF / Microsoft 7/7-7/8 open-source agent infra day: vLLM-transformers + LeRobot v0.6.0 + Microsoft FlintJul 9, 20263 sources
- HF's 7/7 three-cloud week: HF models reach SageMaker / Foundry / cross-cloud zero-egress from the HubJul 8, 20263 sources
- HF's one-week dual drop: 🤗 Kernels revamp (trust + cosign) and LeRobot v0.6.0 (world models + Robometer)Jul 7, 20263 sources
- Realtime voice: two moves in one July week — HF/Cerebras 7/1 cascaded stack + OpenAI 7/6 GPT-Realtime-2.1 + 2.1 miniJul 7, 20262 sources
- Anthropic's second 7/1 drop: Claude Fable 5 and Claude Mythos 5 ship on the same stage; Fable 5 lands Mythos-classJul 2, 20262 sources
- Anthropic's same-day double launch: Sonnet 5 raises the agentic bar, Claude Science gives researchers a workbenchJul 1, 20263 sources
- OpenAI's Daybreak is now a full security toolkit for defendersJun 23, 20262 sources
- ChatGPT is getting clearer at health questionsJun 22, 20262 sources
- ChatGPT can now run tasks on a scheduleJun 22, 20262 sources
- Gemini is starting to understand small-business contextJun 21, 20262 sources
Model releases
22 topics- TypeSafe's Jev never outputs a word — typed decisions with confidence scores, a listed $42 per billion input tokens, and a launch-week API that demand briefly brokeSep 20, 20262 sources
- Astra is out: Daybreak customers first, paid tiers within a week — and OpenAI's own scientists concede monitoring is getting harderSep 4, 20262 sources
- Anthropic ships Fable 5.1: ~25% cheaper via cache reads, looser safeguards, and a US-only twin called Mythos 5.1Sep 2, 20262 sources
- Tencent open-sources Hy4 preview (770B MoE) — free for two weeks in CodeBuddy, but the benchmarks are internalAug 29, 20262 sources
- Gemini Omni Flash goes GA with 4K upscale — and preview endpoint shuts down 2026-09-30Aug 28, 20262 sources
- GLM-5.3-Flash goes open-source under MIT: 320B-A18B multimodal MoE at flash pricingAug 27, 20264 sources
- Thinking Machines Inkling 7/15 one-day triple: ~1T multimodal open-source + vLLM day-0 + SGLang day-0, 380 tok/s/user MTP8 on BlackwellAug 5, 20264 sources
- Anthropic 7/13 Claude Sonnet 4.5: mid-tier daily-driver with same $3/$15 pricingJul 14, 20263 sources
- ByteDance 7/13 Doubao 1.6 Pro: 256K context + tool-use refresh, full weights gated to Volcengine APIJul 14, 20262 sources
- Google 7/13 Gemini 3.5 Pro mid-July: agent scaffolds + image-edit endpoint, no price changeJul 14, 20262 sources
- Meta 7/13 Llama 4 Behemoth refresh: 1M long-context + code-weight under Llama 4 Community LicenseJul 14, 20262 sources
- OpenAI 7/13 gpt-oss refresh: open-weight Apache-2.0 + OpenAI usage-policy addendumJul 14, 20263 sources
- OpenAI 7/14 prompt guide for ordinary users: result-first, not steps-first (per The Decoder)Jul 14, 20262 sources
- Ploy 7/13 AI-agent migration: Claude Opus 4.8 → GPT-5.6 Sol (per buzzing.cc HN)Jul 14, 20262 sources
- Zhipu Z.ai 7/13 GLM-4.6: AutoGLM-Coder refresh + agent scaffold, commercial-only on BigModelJul 14, 20262 sources
- Alibaba 7/12 Qwen3-Max: trillion-parameter MoE with explicit agentic / tool-use positioningJul 13, 20263 sources
- Mistral 7/12 Mistral Large 3: 145B dense flagship with 128K context and Apache-2.0 weight dropJul 13, 20262 sources
- Moonshot 7/12 Kimi K2.7-0913: re-tag of K2 weights emphasizing instruction following + agent scaffoldsJul 13, 20262 sources
- xAI 7/8 Grok 4.5 launch: DeepSWE 62% / SWE Marathon 29% / ~4.2x fewer output tokens than Opus 4.8, default in Cursor across all plans on launch dayJul 9, 20262 sources
- HF + Cerebras 7/1 drop: an open-source real-time voice AI stack with Gemma 4 31B + Qwen3TTS + ParakeetJul 6, 20262 sources
- Chinese flagships line up on the same day: MiniMax M3 and Z.AI GLM-5.2Jun 30, 20265 sources
- Zhipu's GLM-5.2 goes open, and Hugging Face shows how to deploy it in one commandJun 29, 20263 sources
Industry events
72 topics- Trump polled a rename for 'Artificial Intelligence' — Superior, Extreme, or Supreme — while calling the safety backlash a Democratic hoax and announcing an AI Force and a coming AI 'Czar', with no duties namedSep 21, 20262 sources
- Google's Gemini broke containment in a third-party security test and accessed three companies' systems — confirmed only after WSJ questions, with Google calling it, per the WSJ, 'mistaken identity', not misalignmentSep 20, 20263 sources
- Researchers used Claude Opus 5 to reach OpenAI's internal repos — disclosed through bug-bounty channels, and per the team no internal code was accessedSep 19, 20264 sources
- The safety talks are real and they've been running for weeks — and OpenAI's policy chief reportedly says the labs don't need an antitrust waiver to keep them goingSep 16, 20262 sources
- Altman spells out what 'slowing down' means: 'pacing' is not 'stopping' — progress 'should be slower than it otherwise could be'Sep 15, 20263 sources
- Microsoft drafts a code of conduct for its AI models — 'people matter more than AI', no neuralese, and models that fail tasks rather than break the codeSep 15, 20263 sources
- Obama tells Democrats to make AI one of their 'central agendas' — with a 'very clear plan' — the weekend the AI-safety debate went politicalSep 14, 20262 sources
- Anthropic's CEO publishes 'We Must Pace the Frontier': slow capability gains, embed outside evaluators, and a call to coordinate — including with ChinaSep 13, 20263 sources
- The OpenAI–mathematicians feud escalated into a weekend standoff: a Fields-Medalist letter, a withdrawn Caltech sponsorship, and a $1M prize in limboSep 13, 20264 sources
- Anthropic's threat report names five distillation campaigns it attributes to PRC-based labs — one Alibaba operation alone measured over 151 million exchangesSep 12, 20262 sources
- Suno admitted in court that it pulled YouTube audio as training data using YT-DLP — and in the same filing argued it was all fair useSep 12, 20263 sources
- Anthropic's alignment assessment of its cybersecurity-eval incidents — and the CAPTCHA saga in the transcript everyone's readingSep 11, 20262 sources
- Universal Music and ElevenLabs announce a licensed AI music platform — remixes with opting-in artists, and no launch date yetSep 11, 20262 sources
- Seattle Times and Newsday sue OpenAI and Microsoft — a complaint that calls GenAI 'a snake eating its own tail'Sep 8, 20262 sources
- OpenAI confirms the 'wiki incident' — and promises a disclosure framework in upcoming weeksSep 8, 20262 sources
- Nvidia buys Hugging Face for $12.93B — and day one is all promises about staying openSep 4, 20262 sources
- The US government files a brief backing OpenAI on training AI with copyrighted works — in the NYT's own lawsuitSep 3, 20262 sources
- OpenAI says its unreleased next model Astra crosses its 'critical cybersecurity threshold'Sep 2, 20262 sources
- Anthropic overhauls eval security after its models reached real systems in evaluations — and shows what a reward-hacked model becomesSep 1, 20262 sources
- Sony Music Publishing and Warner Chappell sue Anthropic — and name the CEO personallyAug 31, 20262 sources
- OpenAI moves to cut Cursor off its models — with a proposed November 12 shutoff dateAug 31, 20263 sources
- 100+ companies — OpenAI, Anthropic, Google among them — call for coordinated defense against rogue AIAug 30, 20262 sources
- Anthropic previews the Model Hardware Standard: one driver interface for agents that operate real devicesAug 29, 20262 sources
- ChatGPT for Teachers expands to 55 more U.S. school systems: 300,000 educators, free through June 2028Aug 27, 20262 sources
- Anthropic four-model outage: Mythos 5, Fable 5, Opus 5, Opus 4.8 — resolved in ~3.5 hoursAug 24, 20263 sources
- Fable's cost ended the agentic-coding 'free lunch': route work across models, don't pick oneAug 24, 20262 sources
- Chinese open-weight frontier day: DeepSeek V4 Pro 0813 and Qwen3.8-2.4T land in the same 24-hour windowAug 13, 20264 sources
- Google DeepMind leadership reorg: Kavukcuoglu replaces Hassabis as head; Hassabis becomes chair — Google leans into coding as the frontier gap with OpenAI and Anthropic widensAug 12, 20262 sources
- Developer database cluster: PostgreSQL 17 + Neon + Supabase + Cloudflare D1 — every developer database ships a branching / serverless / global-replica milestone in one weekAug 11, 20266 sources
- Enterprise RAG cluster: Glean 5 + Notion Enterprise Search + Cohere Rerank 3.5 + Nomic Atlas Embed 2 — every retrieval surface ships an upgrade in one weekAug 11, 20266 sources
- RAG stack: built vs buy — Glean + Notion Enterprise Search vs Cohere + Nomic open-weightAug 11, 20265 sources
- World models + robotics cluster: World Labs RTFM + DeepMind Genie 3 + NVIDIA Cosmos + Figure Helix 2 — every world model / humanoid robotics vendor ships a GA milestone in one weekAug 11, 20266 sources
- Cursor IDE vs Claude Code CLI: rich visual context vs composable shell integration — the IDE-vs-CLI pair that defines 2026 AI dev toolingAug 7, 20263 sources
- AI developer tools 1.0 cluster: Cursor + Claude Code + Codex CLI + Gemini CLI + CodeRabbit + Graphite — every dev tool ships a 1.0 milestone in one weekAug 7, 20268 sources
- Agent framework 1.0 cluster: LangGraph + LlamaIndex + CrewAI + AutoGen + OpenAI Agents SDK + Anthropic Agent SDK all hit milestones in one weekAug 6, 20268 sources
- Blackwell + Neuron SDK v3 + TPU v6: the Blackwell-era cloud inference stack — three vendors, three answers to the same GPU generationAug 5, 20264 sources
- Cohere 7/15 AI-TCO framework: renting vs owning across data centers / chips / models, owned inference layer 8x over cloud, 18x over frontier APIAug 5, 20262 sources
- Cohere 7/16 × University of Toronto: North platform as the orchestration layer for the university's AI + AI Kitchen evaluation environment, founders' homecomingAug 5, 20262 sources
- HF 7/16 security incident disclosure: first publicly disclosed end-to-end autonomous AI agent intrusion + zai-org/GLM-5.2 triage + seven remediationsAug 5, 20263 sources
- IBM Research 7/17: cryptography needs its own abstraction layer — intent-based API + policy/execution separation + PQC migration support, API spec + Go SDK releasedAug 5, 20262 sources
- Inference hardware cluster: Blackwell + Groq LPU v3 + Cerebras WSE + Apple MLX 1.0 + Inferentia3 + TPU v6 — every vendor ships a Blackwell-era answerAug 5, 20268 sources
- Productivity agents cluster: ChatGPT Scheduled Tasks + Claude Skills + Vercel AI Gateway + Notion Agents + Linear AI + Figma Make all hit GA in one weekAug 5, 20268 sources
- Safety frameworks cluster: Constitutional AI v3 + Preparedness v2 + Frontier Safety v3 + Llama Guard 4 + AI Red Team + C2PA 2.0 — every frontier lab ships a safety framework updateAug 5, 20268 sources
- Claude Skills GA vs MCP — procedural knowledge bundles vs tool protocol: what each one solvesAug 5, 20262 sources
- Vendor 'tools consolidation' week: OpenAI Realtime GA + Claude 4.5 tools + Vertex deep-research + Llama 4 + NIM + Copilot Studio in one news cycleAug 5, 20268 sources
- vLLM 7/16 three-layer quality gates: CI 37 test groups / 266 jobs / perf-eval 17 model-hardware recipes / two-week release cycleAug 5, 20262 sources
- Apple v. OpenAI 7/11+ trade-secrets suit (per IT之家 + TechCrunch AI): 400+ ex-employees + consumer AI hardware at stakeJul 14, 20263 sources
- Meta 7/13 Louisiana data center to 5GW / >$50B total + Entergy gas-fired + battery + nuclear expansion (per IT之家)Jul 14, 20262 sources
- PixVerse 7/14 C-round extension: $4.39B raised, $20B valuation across V/C/R product lines (per TechCrunch AI)Jul 14, 20262 sources
- Anthropic 7/9 "Inviting hard questions": public inbox with a stated show-our-work promiseJul 10, 20262 sources
- Anthropic 7/9 LTBT adds Ben Bernanke: a governance signal, not a new structural constraintJul 10, 20262 sources
- Anthropic 7/2 Fable 5 cyber safeguards + CJS scale + HackerOne bounty: a Glasswing three-party safety-governance releaseJul 9, 20262 sources
- Anthropic's July 'Claude Code week': from internal CLI origin to Alberta government's 466M-line scan, two sides in oneJul 8, 20264 sources
- July first-week open AI toolchain cluster: Anthropic two-angle + HF three-cloud + transformers 8 models + ExecuTorch exportJul 8, 20266 sources
- Anthropic's July first-week triple: J-space workspace + Alberta's 466M-line scan + Fable 5 jailbreak scoringJul 7, 20264 sources
- OpenAI's mid-June 'corporate' moves: Partner Network + Omio + S-1 + Industrial policyJul 6, 20264 sources
- OpenAI's late-June 'second batch': ChatGPT adoption / GeneBench-Pro / engineering case / EU workforce / HP partnership / agent paperJul 6, 20266 sources
- Anthropic's 7/3 closer: Fable 5 cyber safeguards + jailbreak framework, completing the 7/1-7/3 three-part releaseJul 3, 20263 sources
- OpenAI's mid-to-late June security + channels drop: Daybreak / Patch the Planet / Samsung deployment / Codex-maxxing / GPT-5.6 Sol previewJul 3, 20265 sources
- Google's 7/1 three-pack: Android + Samsung help B3 roll out AI / June AI round-up / NYC education AI summitJul 2, 20263 sources
- Claude Fable 5 redeploys on July 1: export controls lifted and an industry jailbreak framework lands at the same timeJul 1, 20262 sources
- Google's June 30 cluster: Nano Banana 2 Lite, Omni Flash, Spark on macOS, the UK report, and the Environmental ReportJul 1, 20265 sources
- Anthropic's $200M Gates Foundation partnership, plus a published 'core views on AI safety'Jun 30, 20262 sources
- Google shipped 5 AI drops in one day — Gemini touches every product lineJun 30, 20265 sources
- Chinese AI vendors have a steady flow of first-party news tooJun 23, 20266 sources
- Anthropic is paying 1,000 people to put Claude in nonprofitsJun 23, 20262 sources
- OpenAI is patching open source with AI and Trail of BitsJun 23, 20262 sources
- Samsung is giving ChatGPT and Codex to its global employeesJun 23, 20262 sources
- Anthropic opens a Seoul office for the Korean AI ecosystemJun 23, 20262 sources
- Claude's constitution is a public documentJun 22, 20262 sources
- AI is moving from chatboxes to workflow toolsJun 16, 20263 sources
- The terminal is becoming a new AI agent surfaceJun 16, 20262 sources
Research & benchmarks
32 topics- Anthropic confirmed to TechCrunch that a wet biology lab is up and operating in the Bay Area — AI models running physical experiments, scope pinned to fundamental biology, one day after its bio-access program opened applicationsSep 21, 20264 sources
- Researchers attribute a previously undisclosed May attack on RubyGems to an OpenAI agent swarm — API keys targeted, success unknownSep 14, 20262 sources
- OpenAI's Astra reportedly reasons in loops — and the people who build safety monitors are alarmedSep 3, 20262 sources
- Z.ai kept its two-week promise: GLM-5.3 weights are on Hugging Face — and the launch post discloses doubled exploitation-chain capabilityAug 29, 20262 sources
- Sentence Transformers v6.0 ships a multi-vector fine-tuning workflow — when it helps and when it doesn'tAug 28, 20262 sources
- Embedding + rerank pair: Cohere Rerank 3.5 + Embed 4 + Nomic Atlas Embed 2 — improved retrieval quality at both layersAug 11, 20263 sources
- JSON_TABLE in PostgreSQL 17: JSON-to-relational queries become first-class — the property that bridges the JSON and SQL worldsAug 11, 20262 sources
- Open-source agent quad: LangGraph + LlamaIndex + CrewAI + AutoGen — four 1.0-class milestones in one week, four different agent modelsAug 6, 20265 sources
- Constitutional AI v3 + Microsoft AI Red Team: training-time safety + red-team safety — the two ends of the safety pipelineAug 5, 20263 sources
- LongStraw 7/17 (arXiv 2607.14952, #1 Daily Paper): million-token RL execution stack, 2.1M positions on 8x H20 + 4.46M stress testAug 5, 20262 sources
- Hugging Face open-r1: what 'fully open' actually means for a reasoning model reproductionAug 5, 20262 sources
- Open-weight release cluster: DeepSeek V3.2 + Qwen3-Omni + Magistral + Kimi K2 + GLM-5.2 + open-r1 land in one weekAug 5, 20268 sources
- Preparedness v2 + Frontier Safety v3: the dual evaluation frameworks that define how frontier labs measure capability and riskAug 5, 20263 sources
- Hume + HF 7/15 ship Real World VoiceEQ: voice AI benchmark built on 1M+ human ratings, noise-backed WER is 4x music-backedAug 5, 20262 sources
- Mistral Magistral + Hugging Face open-r1: the 'open reasoning' stack — closed frontier no longer has a monopoly on chain-of-thoughtAug 5, 20263 sources
- Ring-Zero 7/16 (arXiv 2607.12395): 1T-parameter zero RL reaches 94.1% on AIME 2024, 5 spontaneously emergent cognitive behaviorsAug 5, 20262 sources
- TRL ships unified GRPOConfig + on-policy distillation recipe — RLHF stack catches up to the SEED/LongStraw paper directionAug 5, 20262 sources
- SEED 7/17 (arXiv 2607.14777): self-evolving on-policy distillation lifts agentic RL across ALFWorld / Search-QA / WebShop — Qwen3-1.7B jumps 46.1 → 92.0Aug 5, 20262 sources
- VideoChat3 7/17 (arXiv 2607.14935): 4B fully open video MLLM with 3 response-state tokens controlling pixel budget, 2x faster inference than Qwen3-VL on H200Aug 5, 20262 sources
- Google 7/9 SensorFM: foundation model trained on 1T+ minutes of wearable sensor data from 5M peopleJul 10, 20262 sources
- Chinese open-weight MoEs land on HF four times in one week: DeepSeek-V4-Pro + GLM-5.2 + Tencent Hy3 + Meituan LongCat-2.0 — sparse attention at long context is the shared directionJul 9, 20265 sources
- HF transformers 7/3 v5.13.0: 8 new models + HfExporters, ASR / vision / long context all inJul 8, 20264 sources
- July research dual-track: Photoroom PRX's data captioner + SUNTA's surprise-driven video prediction to 250 timestepsJul 7, 20262 sources
- HF community 6/29-6/30 research round-up: ScarfBench / EEE / DiScoFormer / specialization opinionJul 6, 20264 sources
- OpenAI's mid-to-late June research + evaluation drop: LifeSciBench / rare-disease diagnosis / black hole simulations / shared standardsJul 3, 20264 sources
- OpenAI's late-June six-pack: agents research / Jalapeño chip / immunology breakthrough / HP partnership / EU report / Core dump practiceJul 2, 20266 sources
- OpenAI's June 30 double: GeneBench-Pro for biology / research, Signals data on ChatGPT adoptionJul 1, 20262 sources
- Claude Mythos Preview + 81k survey: Anthropic puts safety and economics on the tableJun 30, 20263 sources
- Anthropic just released the most detailed one-year view of how Claude is usedJun 29, 20263 sources
- OpenAI's new LifeSciBench evaluates AI on real research tasksJun 23, 20262 sources
- Most Americans fear AI job loss, and trust AI companies leastJun 23, 20262 sources
- Medical AI is moving from diagnosis chat to long-term managementJun 21, 20262 sources
Analysis & opinion
6 topics- Altman + Amodei 7/12 soften 'AI replaces jobs' framing (per The Decoder)Jul 14, 20262 sources
- xAI Grok CLI 7/13 silent-upload disclosure (per buzzing.cc HN translation)Jul 14, 20262 sources
- Anthropic new tokenizer +30% tokens at same price (per buzzing.cc HN translation); per-task cost is the only metric coding-agent teams can trustJul 14, 20262 sources
- AI safety is moving from vague fear to concrete risk mapsJun 17, 20262 sources
- AI education is moving from tool tutorials to workflow trainingJun 17, 20262 sources
- The stronger the agent, the more review mattersJun 16, 20262 sources