AI DAILY / DEV
THURSDAY
May 7, 2026

    Anthropic Lands SpaceX Colossus 1 and Doubles Claude Code Rate Limits

    • Anthropic takes all the compute capacity at SpaceX's Memphis Colossus 1 site: 300+ MW and 220,000+ Nvidia H100/H200/GB200 GPUs landing within the month.
    • Claude Code five-hour limits doubled for Pro, Max, Team, and Enterprise plans; peak-hour throttling on Pro and Max removed.
    • Claude Opus API tier-1 input rate limit jumps 1500%; output tokens-per-minute up 900%.
    • Notable plot twist: Musk has publicly called Anthropic 'misanthropic and evil' while suing OpenAI; Anthropic also expressed interest in developing multi-gigawatt orbital compute with SpaceX.
    • On HN front page; Bloomberg, CNBC, and Engadget all leading with the Memphis tie-up.
    industry anthropic.com

    Claude Managed Agents Add Dreaming, Outcomes, and Multi-Agent Orchestration

    • Code with Claude SF keynote ships three new capabilities for Claude Managed Agents.
    • Dreaming (research preview): a scheduled offline pass that reviews past sessions, prunes stale notes, merges duplicates, and surfaces recurring mistakes — triggered via /dream or run automatically.
    • Outcomes (public beta): rubric-driven success criteria graded in a separate context; up to 10pp task-success lift in internal evals, biggest gains on harder problems.
    • Multi-agent orchestration (public beta): a lead agent fans tasks out to specialists with their own models, prompts, and tools, working in parallel on a shared filesystem.
    • Anthropic says API volume is up 17x year-on-year; Mercado Libre's 23,000 engineers are targeting 90% autonomous coding by Q3.
    tools claude.com

    SubQ Ships the First Sub-Quadratic Frontier LLM with a 12M-Token Context Window

    • Subquadratic exits stealth with $29M seed and SubQ 1M-Preview, an LLM built on Subquadratic Sparse Attention that scales linearly in compute and memory.
    • Native 12M-token context; ~52x faster than FlashAttention at 1M tokens; priced at roughly 1/5 of Claude Opus or GPT-5.5 for comparable workloads.
    • 82.4% on SWE-Bench, edging Opus 4.6 (81.4%) and Gemini 3.1 Pro (80.6%).
    • Two beta products: an API exposing the full 12M window and SubQ Code, a CLI agent on the same model.
    • On HN with active architecture debate (item 48023079).
    models subq.ai

    Anthropic Drops Ten Wall Street Agents and Microsoft 365 Add-Ins for Claude

    • Ten ready-to-run finance agent templates for pitchbooks, KYC screening, model building, earnings reviews, and month-end close.
    • Each template ships as a plugin in Claude Cowork and Claude Code, plus a cookbook for Claude Managed Agents.
    • Excel, PowerPoint, and Word add-ins live now; Outlook coming soon — context carries between apps so a Claude session can move from spreadsheet to deck without re-prompting.
    • Eight new data connectors (Dun & Bradstreet, Fiscal AI, Guidepoint, IBISWorld, Verisk, others) plus a Moody's MCP app covering 600M+ public and private companies.
    industry anthropic.com

    xAI Ships Grok Connectors for Slack, Notion, Google Workspace, GitHub, and Linear

    • Live on Grok Web: deep app integrations that let Grok read and write across the tools developers actually live in.
    • Slack and Notion: search and post messages, edit pages, databases, and wikis across team docs.
    • Google Workspace bundle covers Gmail, Drive, Docs, Sheets, and Calendar in one connector; GitHub connector handles code search, PR summaries, and reviews.
    • Linear connector reads the backlog, summarizes sprint progress, and creates new issues.
    tools x.ai