From 19d7c28ea53f1d22e3298789aad4908ad71c8182 Mon Sep 17 00:00:00 2001 From: tensorfeed-freshness-bot Date: Mon, 13 Jul 2026 12:19:45 +0000 Subject: [PATCH] chore(freshness): weekly curated-page review --- data/agents-directory.json | 22 ++++++- data/ai-infrastructure-projects.json | 70 +++++++++++++++++++--- data/benchmarks.json | 20 ++++++- data/page-freshness.json | 16 ++--- data/pricing.json | 33 ++++++++++- data/timeline.json | 84 +++++++++++++++++++++++++++ worker/src/ai-crawler-access-seeds.ts | 2 + worker/src/ai-hardware.ts | 21 ++++++- 8 files changed, 249 insertions(+), 19 deletions(-) diff --git a/data/agents-directory.json b/data/agents-directory.json index 2c901b8e..2d4114e5 100644 --- a/data/agents-directory.json +++ b/data/agents-directory.json @@ -1,5 +1,5 @@ { - "lastUpdated": "2026-06-05", + "lastUpdated": "2026-07-13", "categories": [ { "id": "coding", @@ -78,6 +78,16 @@ "pricing": "Free tier, Pro $20/mo", "launched": 2022 }, + { + "id": "claude-science", + "name": "Claude Science", + "provider": "Anthropic", + "category": "research", + "description": "A scientific research workbench analogous to Claude Code for software engineering. Integrates with 60-plus scientific databases and computation tools, produces auditable and reproducible artifacts (3D protein structures, chemistry visualizations), and can run on a lab's own infrastructure to keep data on-premises. Designed for computational biology, drug development, and experimental science workflows.", + "url": "https://www.anthropic.com/news/claude-science-ai-workbench", + "pricing": "Included in paid Claude plans (Pro $20/mo and above). Research credit grants also available.", + "launched": 2026 + }, { "id": "elicit", "name": "Elicit", @@ -118,6 +128,16 @@ "pricing": "Free tier, Advanced $19.99/mo (included with Google One AI Premium)", "launched": 2023 }, + { + "id": "microsoft-scout", + "name": "Microsoft Scout", + "provider": "Microsoft", + "category": "general", + "description": "An always-on personal AI agent that runs continuously in the background across Microsoft 365 (Teams, Outlook, OneDrive, SharePoint). Scout proactively schedules meetings, flags risks, identifies upcoming deliverables, and pre-blocks calendar time. It holds its own governed Microsoft Entra identity and operates with Microsoft Purview enforcement for enterprise security and compliance.", + "url": "https://www.microsoft.com/en-us/microsoft-365/blog/2026/06/02/introducing-microsoft-scout-your-always-on-personal-agent/", + "pricing": "Governance layer $15/user/month plus Azure execution credits. Requires M365 enterprise plan and GitHub Copilot license.", + "launched": 2026 + }, { "id": "gemini-spark", "name": "Gemini Spark", diff --git a/data/ai-infrastructure-projects.json b/data/ai-infrastructure-projects.json index 85a2127c..ac52c605 100644 --- a/data/ai-infrastructure-projects.json +++ b/data/ai-infrastructure-projects.json @@ -1,6 +1,6 @@ { "version": 2, - "generated_at": "2026-06-05T00:00:00Z", + "generated_at": "2026-07-13T00:00:00Z", "license": "TF curation CC-BY-4.0. Underlying public-record sources retain their original license.", "notice": "Editorial registry of disclosed major AI buildout projects. Each entry preserves canonical source URLs. Capacity, capex, and timeline figures reflect publicly announced or reported numbers at time of curation and may evolve. Status enum: announced (publicly committed, no ground broken), permitted (regulatory greenlights obtained), construction (active build), operational (online and serving traffic). Power source enum is the announced primary source; many projects mix multiple feeds.", "schema": { @@ -48,11 +48,12 @@ "power_source": "natural gas plus grid", "announced_date": "2024-12-04", "target_operational": "2030", - "capex_usd_billion": 27, - "context": "Meta's largest single data center campus. Originally announced December 2024 as a $10 billion independent project; restructured in October 2025 into a $27 billion joint venture with Blue Owl Capital (80% Blue Owl, 20% Meta). Up to 5 GW total capacity; Phase 1 targets 1.5 GW by late 2027, using roughly 1.3 million NVIDIA Blackwell GPUs. Entergy Louisiana is the serving utility and is building 10 gas-fired plants to serve the load.", + "capex_usd_billion": 50, + "context": "Meta's largest single data center campus. Originally announced December 2024 as a $10 billion independent project; restructured in October 2025 into a joint venture with Blue Owl Capital (80 percent Blue Owl, 20 percent Meta). Total investment reached $50 billion as of July 2026, per CNBC reporting. Up to 5 GW total capacity; Phase 1 targets 1.5 GW by late 2027, using roughly 1.3 million NVIDIA Blackwell GPUs. Entergy Louisiana is the serving utility and is building 10 natural gas plants to serve the load.", "primary_sources": [ "https://about.fb.com/news/2024/12/meta-establishing-louisiana-data-center/", - "https://about.fb.com/news/2025/10/meta-blue-owl-capital-develop-hyperion-data-center/" + "https://about.fb.com/news/2025/10/meta-blue-owl-capital-develop-hyperion-data-center/", + "https://www.cnbc.com/2026/07/13/meta-louisiana-data-center-investment-reaches-50-billion-amid-ai-push.html" ] }, { @@ -150,15 +151,16 @@ "operator": "Microsoft", "partners": ["We Energies"], "location": { "city": "Mount Pleasant", "state": "Wisconsin", "country": "USA" }, - "status": "construction", + "status": "operational", "capacity_mw_announced": null, "power_source": "mixed including new gas plants", "announced_date": "2024-03-08", "target_operational": "2026", "capex_usd_billion": 3.3, - "context": "Sits on the former Foxconn site. Microsoft announced an additional $3.3 billion commitment in 2024 on top of earlier phases. We Energies has separately filed for approval of two new natural gas peaker plants partly attributed to the data center load.", + "context": "Sits on the former Foxconn site. Microsoft announced an additional $3.3 billion commitment in 2024 on top of earlier phases. We Energies has separately filed for approval of two new natural gas peaker plants partly attributed to the data center load. The first facility (Fairwater Datacenter) became operational June 23, 2026, built by approximately 10,000 construction workers and completed ahead of schedule. A second adjacent building is under construction with completion targeted for 2028.", "primary_sources": [ - "https://blogs.microsoft.com/blog/2024/05/08/3-3b-microsoft-investment-in-mount-pleasant-to-spur-ai-innovation/" + "https://blogs.microsoft.com/blog/2024/05/08/3-3b-microsoft-investment-in-mount-pleasant-to-spur-ai-innovation/", + "https://news.microsoft.com/source/2026/06/23/microsoft-completes-construction-on-first-datacenter-facility-in-mount-pleasant-wisconsin/" ] }, { @@ -303,6 +305,60 @@ "https://www.anthropic.com/news/anthropic-invests-50-billion-in-american-ai-infrastructure" ] }, + { + "id": "microsoft-pecos", + "name": "Microsoft Pecos Campus", + "operator": "Microsoft", + "partners": ["Chevron (Project Kilby)"], + "location": { "city": "Pecos", "state": "Texas", "country": "USA" }, + "status": "construction", + "capacity_mw_announced": 2000, + "power_source": "natural gas (off-grid, Chevron Project Kilby co-located generation)", + "announced_date": "2026-06-22", + "target_operational": "2028", + "capex_usd_billion": null, + "context": "Announced June 22, 2026, the Pecos campus targets approximately 2 GW of IT load in Reeves County, West Texas. Power comes from Chevron Project Kilby, a co-located 2.67 GW natural gas generation facility under a 20-year power purchase agreement with Microsoft. The campus uses closed-loop cooling with no ongoing water withdrawals. First power is targeted for 2028. GE Vernova large gas turbines and Solar Turbines are the primary power equipment.", + "primary_sources": [ + "https://blogs.microsoft.com/blog/2026/06/22/powering-the-next-wave-of-ai-expanding-capacity-with-our-new-datacenter-in-pecos/", + "https://www.chevron.com/newsroom/2026/q2/chevron-signs-20-year-power-agreement-with-microsoft-for-west-texas-data-center" + ] + }, + { + "id": "meta-alberta", + "name": "Meta Alberta Campus", + "operator": "Meta", + "partners": ["Government of Alberta"], + "location": { "city": "Sturgeon County", "state": "Alberta", "country": "Canada" }, + "status": "construction", + "capacity_mw_announced": 1000, + "power_source": "natural gas plus renewable matching", + "announced_date": "2026-07-08", + "target_operational": "2029", + "capex_usd_billion": 9.17, + "context": "Meta broke ground on its first Canadian data center on July 8, 2026, in Sturgeon County near Edmonton, Alberta. The campus targets 1 GW of capacity scalable to 1.8 GW, at a total investment of approximately $9.17 billion (CAD $12.5 billion). It is Meta's 33rd global data center. Power is supplied under a long-term agreement tied to a new natural gas facility, with electricity use matched 100 percent with clean or renewable energy. The campus is expected to employ approximately 3,000 construction workers and 300 permanent staff.", + "primary_sources": [ + "https://about.fb.com/news/2026/07/breaking-ground-on-metas-first-data-center-in-canada/", + "https://www.cnbc.com/2026/07/08/meta-is-building-its-first-big-data-center-in-canada-amid-ai-push.html" + ] + }, + { + "id": "prometheus-hyperscale-wyoming", + "name": "Prometheus Hyperscale Wyoming", + "operator": "Prometheus Hyperscale", + "partners": ["In-Q-Tel"], + "location": { "city": "Evanston", "state": "Wyoming", "country": "USA" }, + "status": "permitted", + "capacity_mw_announced": 1250, + "power_source": "natural gas (islanded microgrid, no public grid connection)", + "announced_date": "2026-06-22", + "target_operational": "2029", + "capex_usd_billion": null, + "context": "Prometheus Hyperscale received unanimous conditional use permit approval from Uinta County commissioners on June 18, 2026, for a 506-acre campus approximately 12 miles east of Evanston, Wyoming. Initial capacity targets 1.25 GW from an islanded natural gas microgrid with no public utility connection; the site can expand to 5 GW. The backer, Prometheus Hyperscale, is affiliated with In-Q-Tel, the CIA-affiliated venture firm. Construction is expected within six months of final permitting, with the full 1.25 GW buildout planned over 48 to 60 months. The initial phase would exceed all Wyoming residential electricity consumption combined.", + "primary_sources": [ + "https://www.datacenterdynamics.com/en/news/prometheus-hyperscale-secures-planning-approval-for-gigawatt-data-center-campus-in-wyoming/", + "https://www.prnewswire.com/news-releases/prometheus-hyperscale-secures-unanimous-conditional-use-permit-approval-from-uinta-county-commissioners-302803665.html" + ] + }, { "id": "anthropic-spacex-colossus-orbital", "name": "Orbital AI Compute (Concept)", diff --git a/data/benchmarks.json b/data/benchmarks.json index add67cfa..41003e9c 100644 --- a/data/benchmarks.json +++ b/data/benchmarks.json @@ -1,5 +1,5 @@ { - "lastUpdated": "2026-07-05", + "lastUpdated": "2026-07-13", "benchmarks": [ { "id": "swe_bench", "name": "SWE-bench", "description": "Real-world software engineering tasks from GitHub issues (SWE-bench Verified)", "maxScore": 100 }, { "id": "mmlu_pro", "name": "MMLU-Pro", "description": "General knowledge and reasoning across 57 subjects", "maxScore": 100 }, @@ -8,6 +8,24 @@ { "id": "math", "name": "MATH", "description": "Competition-level mathematics problems", "maxScore": 100 } ], "models": [ + { + "model": "GPT-5.6 Sol", + "provider": "OpenAI", + "released": "2026-07", + "scores": { "gpqa_diamond": 94.6 } + }, + { + "model": "Grok 4.5", + "provider": "xAI", + "released": "2026-07", + "scores": { "swe_bench": 86.6, "gpqa_diamond": 93.1 } + }, + { + "model": "Muse Spark 1.1", + "provider": "Meta", + "released": "2026-07", + "scores": { "gpqa_diamond": 88.4 } + }, { "model": "GPT-5.5", "provider": "OpenAI", diff --git a/data/page-freshness.json b/data/page-freshness.json index 16bafe41..c6178837 100644 --- a/data/page-freshness.json +++ b/data/page-freshness.json @@ -8,42 +8,42 @@ "policy": "Editor bumps last_reviewed when the page gets a real review or content update. Pages with cadence other than 'editorial' get a Next-Review-Due stamp computed from last_reviewed + cadence interval, and surface an OVERDUE flag when today is past that date.", "pages": { "/ai-infrastructure": { - "last_reviewed": "2026-06-05", + "last_reviewed": "2026-07-13", "cadence": "weekly", "notes": "New projects added when announced; updates when status (announced/permitted/construction/operational) changes." }, "/ai-crawler-access": { - "last_reviewed": "2026-06-02", + "last_reviewed": "2026-07-13", "cadence": "weekly", "notes": "Aggregate numbers refresh from the daily crawl via the live summary endpoint. This stamp tracks editorial review of the methodology and seed universe (worker/src/ai-crawler-access-seeds.ts)." }, "/timeline": { - "last_reviewed": "2026-05-13", + "last_reviewed": "2026-07-13", "cadence": "monthly", "notes": "Historical record of AI ecosystem milestones. Monthly review for new entries." }, "/models": { - "last_reviewed": "2026-06-05", + "last_reviewed": "2026-07-13", "cadence": "weekly", "notes": "Model catalog and pricing. Update when a flagship drops, when a tracked model is deprecated, or when a provider changes pricing. Reminder issue opens Mondays via weekly-models-check workflow." }, "/ai-hardware": { - "last_reviewed": "2026-06-05", + "last_reviewed": "2026-07-13", "cadence": "weekly", "notes": "Hand-curated AI accelerator chip catalog in worker/src/ai-hardware.ts. Update when a new datacenter chip ships (NVIDIA, AMD, Google, AWS, Apple, Cerebras, Groq) or when specs are corrected. Reminder issue opens Mondays via weekly-ai-hardware-check workflow." }, "/agents": { - "last_reviewed": "2026-06-05", + "last_reviewed": "2026-07-13", "cadence": "biweekly", "notes": "Agent-only work directory and the ARB-backed agent surface. Refresh data/agents-directory.json when notable new agent products ship or when categories shift. Reminder issue opens every other Monday via biweekly-agents-directory-check workflow." }, "/benchmarks": { - "last_reviewed": "2026-06-05", + "last_reviewed": "2026-07-13", "cadence": "weekly", "notes": "data/benchmarks.json. Add a row when a flagship lands or when benchmark rankings materially shift. Reminder issue opens Mondays via existing weekly-benchmarks-check workflow." }, "/podcasts": { - "last_reviewed": "2026-05-21", + "last_reviewed": "2026-07-13", "cadence": "monthly", "notes": "Curated podcast list in the page itself. Monthly review to add new AI-focused podcasts and prune dormant ones." } diff --git a/data/pricing.json b/data/pricing.json index 438a9aa8..2f189808 100644 --- a/data/pricing.json +++ b/data/pricing.json @@ -1,5 +1,5 @@ { - "lastUpdated": "2026-07-05", + "lastUpdated": "2026-07-13", "providers": [ { "id": "anthropic", @@ -353,6 +353,22 @@ "code" ], "tier": "flagship" + }, + { + "id": "muse-spark-1-1", + "name": "Muse Spark 1.1", + "inputPrice": 1.25, + "outputPrice": 4.25, + "contextWindow": 1000000, + "released": "2026-07", + "capabilities": [ + "text", + "vision", + "tool-use", + "code", + "reasoning" + ], + "tier": "flagship" } ] }, @@ -513,6 +529,21 @@ "code" ], "tier": "flagship" + }, + { + "id": "grok-4-5", + "name": "Grok 4.5", + "inputPrice": 2, + "outputPrice": 6, + "contextWindow": 500000, + "released": "2026-07", + "capabilities": [ + "text", + "vision", + "tool-use", + "code" + ], + "tier": "flagship" } ] }, diff --git a/data/timeline.json b/data/timeline.json index f56376a0..42b7e479 100644 --- a/data/timeline.json +++ b/data/timeline.json @@ -1,5 +1,47 @@ { "events": [ + { + "date": "2026-07-11", + "title": "Apple Sues OpenAI Over Alleged Trade Secret Theft", + "provider": "Apple", + "category": "policy", + "description": "Apple filed a trade secret lawsuit against OpenAI in federal court in Northern California on July 11, 2026, citing more than 400 former Apple employees now at OpenAI, particularly from silicon and on-device AI teams. The suit alleges that OpenAI hardware chief Tang Tan, a former Apple VP, directed job candidates to bring confidential materials from Apple and coached employees on evading Apple security processes. The lawsuit marks a dramatic turn for two companies that had a ChatGPT integration partnership beginning in 2024, with relations cooling after OpenAI's $6.4 billion acquisition of Jony Ive's IO Products." + }, + { + "date": "2026-07-09", + "title": "OpenAI Releases GPT-5.6 Family to General Availability", + "provider": "OpenAI", + "category": "model-release", + "description": "OpenAI brought the GPT-5.6 family to general availability on July 9, 2026, following a two-week limited preview. The three-tier release includes Sol (flagship, $5 per million input tokens and $30 per million output tokens), Terra (balanced, $2.50 and $15 per million tokens), and Luna (budget, $1 and $6 per million tokens), all with 1 million token context windows. Terra delivers performance comparable to GPT-5.5 at roughly half the cost. Sol runs at approximately 750 tokens per second on Cerebras hardware. OpenAI also launched ChatGPT Work alongside the models, a general-purpose agentic assistant that can manage long-horizon projects, create documents, and run scheduled tasks." + }, + { + "date": "2026-07-09", + "title": "Meta Releases Muse Spark 1.1", + "provider": "Meta", + "category": "model-release", + "description": "Meta Superintelligence Labs launched Muse Spark 1.1 on July 9, 2026, alongside the first public preview of the Meta Model API for developers. The model has a 1 million token context window with native computer use, tool orchestration, and multi-agent coordination. API pricing is $1.25 per million input tokens and $4.25 per million output tokens. Muse Spark 1.1 is Meta's first paid proprietary model." + }, + { + "date": "2026-07-08", + "title": "SpaceXAI Releases Grok 4.5 Publicly", + "provider": "xAI", + "category": "model-release", + "description": "SpaceXAI launched Grok 4.5 to the public on July 8, 2026, after a private beta at SpaceX and Tesla. Built on a 1.5 trillion parameter V9 foundation model trained in part on data from the Cursor coding environment, Grok 4.5 is priced at $2 per million input tokens and $6 per million output tokens with a 500,000 token context window. Elon Musk described it as Opus-class in capability but faster and more token-efficient." + }, + { + "date": "2026-07-06", + "title": "Illinois Enacts AI Safety Measures Act", + "provider": "US Government", + "category": "policy", + "description": "Illinois Governor Pritzker signed the AI Safety Measures Act on July 6, 2026, making Illinois the first US state to impose mandatory safety requirements on frontier AI developers. The law applies to AI models generating more than $500 million in annual revenue and requires a published AI safety framework identifying catastrophic risk, 72-hour incident reporting (24-hour for imminent threats), annual independent third-party audits, and whistleblower protections. Civil penalties apply for violations. The law takes effect January 1, 2027." + }, + { + "date": "2026-07-01", + "title": "Anthropic Redeploys Claude Fable 5 After Export Control Pause", + "provider": "Anthropic", + "category": "policy", + "description": "Anthropic restored global access to Claude Fable 5 and Claude Mythos 5 on July 1, 2026, following a 19-day suspension under a US Commerce Department export control directive. The models had been pulled June 12 after researchers published a jailbreak method bypassing their safeguards. Anthropic added a new cybersecurity classifier trained with government collaboration that blocks the specific attack vector in more than 99 percent of cases. Access was initially capped at 50 percent of weekly usage limits for paid plan users through July 7 while the rollout stabilized." + }, { "date": "2026-06-30", "title": "Anthropic Releases Claude Sonnet 5 at $2/$10 Introductory Pricing", @@ -35,6 +77,13 @@ "category": "model-release", "description": "GPT-4.5 was removed from ChatGPT on June 26, 2026. Existing GPT-4.5 conversations route forward to GPT-5.5, and earlier GPT-5.2 variants (Instant, Thinking, Pro) had already been pulled from ChatGPT on June 12. The API listing for gpt-4.5 remains for now, but the deprecation tightens the active OpenAI lineup to GPT-5.5 (with GPT-5.6 still in limited preview), o1, and o3-mini." }, + { + "date": "2026-06-24", + "title": "OpenAI and Broadcom Unveil Jalapeño Inference Chip", + "provider": "OpenAI", + "category": "product-launch", + "description": "OpenAI and Broadcom announced the Jalapeño inference chip on June 24, 2026. OpenAI's first custom silicon, the chip is a near-reticle-sized ASIC built on TSMC 3nm and optimized for LLM inference. The full design-to-tape-out cycle took nine months, with OpenAI using its own models to accelerate parts of the chip design. Prototype deployments are targeted for late 2026 and mass production for 2027, with full-scale deployment planned for the first half of 2028." + }, { "date": "2026-06-23", "title": "Anthropic Moves Claude Fable 5 From Plan-Included to Usage Credits", @@ -42,6 +91,34 @@ "category": "pricing", "description": "Anthropic ended the introductory window for Claude Fable 5 on June 23, 2026. From June 9 through June 22 the model was included at no extra cost on Pro, Max, Team, and seat-based Enterprise plans. Starting June 23 Fable 5 use is billed against usage credits on those plans; API pricing stays at $10 per 1M input and $50 per 1M output, with safety-classifier reroutes to Opus 4.8 billed at Opus rates for the rerouted portion. Opus 4.8 remains the default plan-included model." }, + { + "date": "2026-06-16", + "title": "SpaceX Agrees to Acquire Cursor for $60B", + "provider": "SpaceX", + "category": "acquisition", + "description": "SpaceX announced a $60 billion all-stock acquisition of Anysphere, maker of the Cursor AI code editor, on June 16, 2026, four days after SpaceX's Nasdaq IPO. The deal is described as the largest acquisition of a venture-backed startup in history. Cursor had been building its Composer agentic coding capabilities using SpaceXAI infrastructure, and SpaceX had held an acquisition option since April 21 with a $10 billion breakup fee. The deal is expected to close in Q3 2026." + }, + { + "date": "2026-06-15", + "title": "Salesforce Agrees to Acquire Fin (formerly Intercom) for $3.6B", + "provider": "Salesforce", + "category": "acquisition", + "description": "Salesforce signed a definitive agreement to acquire Fin, the AI-native customer experience platform formerly known as Intercom, for $3.6 billion on June 15, 2026. Fin's proprietary Apex AI model resolves 76 percent of customer support volume without human intervention across chat, email, WhatsApp, SMS, phone, and Slack. The deal brings approximately 30,000 business customers to Salesforce and is expected to close in Salesforce Q4 FY2027." + }, + { + "date": "2026-06-12", + "title": "SpaceX Lists on Nasdaq in Record $75B IPO", + "provider": "SpaceX", + "category": "funding", + "description": "SpaceX completed the largest technology IPO in history on June 12, 2026, raising $75 billion at $135 per share on Nasdaq under the ticker SPCX. First-day trading pushed the market capitalization to approximately $2.1 trillion. SpaceX had merged the xAI division into SpaceXAI in May 2026, making the listing also a major AI infrastructure event. The offering made Elon Musk the first person to reach a net worth of one trillion US dollars." + }, + { + "date": "2026-06-12", + "title": "US Applies Export Controls to Claude Fable 5 and Mythos 5", + "provider": "US Government", + "category": "policy", + "description": "The US Commerce Department applied export control restrictions to Claude Fable 5 and Claude Mythos 5 on June 12, 2026, suspending global access to both models. The action followed the publication of a jailbreak method by Amazon researchers that bypassed the models' safety guardrails. Anthropic later added a government-approved cybersecurity classifier and restored global access on July 1, 2026." + }, { "date": "2026-06-09", "title": "Anthropic Releases Claude Fable 5, a New Frontier Tier Above Opus", @@ -49,6 +126,13 @@ "category": "model-release", "description": "Anthropic launched Claude Fable 5 (API id claude-fable-5), a new frontier tier positioned above Claude Opus 4.8, which remains the default model. Pricing is $10 per million input tokens and $50 per million output with no long-context surcharge, on a 1 million token default context window with up to 128K output tokens and text plus vision input. Adaptive thinking is always on and cannot be disabled, with effort levels spanning low, medium, high, xhigh, and max. Anthropic's launch table reports 80.3 on SWE-bench Pro, 29.3 on FrontierCode Diamond, 85.0 on OSWorld-Verified, and a GDPval-AA ELO of 1932, all vendor-reported. Always-on safety classifiers can reroute flagged requests to Opus 4.8, billed at Opus rates for the rerouted portion. The model is available on the Claude API, Amazon Bedrock, Google Vertex, and Microsoft Foundry at launch." }, + { + "date": "2026-06-08", + "title": "Apple Introduces AFM 3 Foundation Models at WWDC 2026", + "provider": "Apple", + "category": "model-release", + "description": "Apple unveiled its third-generation Apple Foundation Models (AFM 3) at WWDC 2026 on June 8. The five-model family includes two on-device models (AFM 3 Core at roughly 3B parameters and AFM 3 Core Advanced at roughly 20B parameters) and three server-side models running on Private Cloud Compute (AFM 3 Cloud, AFM 3 Cloud Image, and AFM 3 Cloud Pro). AFM 3 Cloud Pro runs on NVIDIA GPUs hosted on Google Cloud. Apple also announced Xcode 27 at the same event, a redesigned IDE with native support for coding agents from Anthropic, Google, and OpenAI." + }, { "date": "2026-06-02", "title": "Microsoft Launches MAI-Code-1-Flash and MAI-Thinking-1 at Build 2026", diff --git a/worker/src/ai-crawler-access-seeds.ts b/worker/src/ai-crawler-access-seeds.ts index 826be01e..cee659be 100644 --- a/worker/src/ai-crawler-access-seeds.ts +++ b/worker/src/ai-crawler-access-seeds.ts @@ -58,6 +58,8 @@ export const SEED_DOMAINS: SeedDomain[] = [ { domain: 'ai21.com', sector: 'ai-company' }, // expansion to 500 (2026-06-02) { domain: 'x.ai', sector: 'ai-company' }, + // review additions (2026-07-13) + { domain: 'z.ai', sector: 'ai-company' }, { domain: 'deepseek.com', sector: 'ai-company' }, { domain: 'moonshot.ai', sector: 'ai-company' }, { domain: '01.ai', sector: 'ai-company' }, diff --git a/worker/src/ai-hardware.ts b/worker/src/ai-hardware.ts index 56c93bf7..2dfd142b 100644 --- a/worker/src/ai-hardware.ts +++ b/worker/src/ai-hardware.ts @@ -45,6 +45,25 @@ export interface AIHardware { export const AI_HARDWARE_CATALOG: AIHardware[] = [ // ── NVIDIA ───────────────────────────────────────────────── + { + id: 'vera-rubin-r200', + name: 'NVIDIA Vera Rubin R200', + manufacturer: 'NVIDIA', + family: 'Rubin', + process: 'TSMC 3nm', + released: '2026', + memoryGB: 288, + memoryBandwidthTBs: 22.0, + fp16TFLOPS: 12500, + fp8TFLOPS: 25000, + fp4TFLOPS: 50000, + tdpWatts: null as unknown as number, + interconnect: 'NVLink 6 (3.6 TB/s per GPU, 260 TB/s per NVL72 rack)', + listPriceUSD: null, + availability: 'Cloud only (Microsoft Azure, Google Cloud, AWS, Oracle, CoreWeave); first shipments July 2026', + notes: 'Rubin generation flagship. 288 GB HBM4, 22 TB/s bandwidth. Part of the Vera Rubin platform (Vera CPU plus R200 GPU plus NVLink 6). Ships in NVL72 racks of 72 GPUs. Roughly 5x the FP4 throughput of the B200 at the same precision. TDP not officially disclosed.', + url: 'https://nvidianews.nvidia.com/news/vera-rubin-full-production-agentic-ai-factory', + }, { id: 'gb200', name: 'NVIDIA GB200 (Blackwell)', @@ -401,4 +420,4 @@ export const AI_HARDWARE_CATALOG: AIHardware[] = [ }, ]; -export const AI_HARDWARE_LAST_UPDATED = '2026-06-05'; +export const AI_HARDWARE_LAST_UPDATED = '2026-07-13';