Anthropic Launches Free Claude for Teachers Across All 50 States
Anthropic rolled out Claude for Teachers, a free tier built specifically for verified US K-12 educators. It bundles premium Claude access, a library of purpose-built teaching skills, and a direct connection to evidence-based curricula mapped to academic standards in all 50 states. Teachers who sign up by June 30, 2027 lock in a full year of access.
It’s also a land grab: Anthropic, OpenAI, and Google are all racing to put their assistants in front of teachers. Anthropic is pairing the launch with a formal pilot in the Detroit Public Schools Community District, studying the effect on educator wellbeing and day-to-day practice rather than just adoption numbers — a bid to answer the “does this actually help teachers?” question head-on.
Claude Science: An AI Workbench Built for Researchers
Anthropic introduced Claude Science, an app aimed squarely at working scientists. Instead of a general chat window, it integrates the tools and packages researchers actually use, produces auditable artifacts so results can be traced and reproduced, and offers flexible access to computing resources for heavier analysis.
The framing matters: reproducibility and provenance are the two things that make or break AI in a lab. By emphasizing artifacts you can audit rather than answers you have to trust, Anthropic is pitching Claude Science as a research collaborator that fits scientific rigor instead of fighting it.
Anthropic Publishes a CISO’s Guide to Agentic AI
On July 17, Anthropic published “Zero risk isn’t the job: a CISO’s guide to agentic AI.” The post reframes the security conversation around agents: the goal isn’t to eliminate every risk before deploying, but to manage it deliberately as autonomous systems take on real work.
It’s a telling piece of positioning. As Claude Code, subagents, and long-running agents move into production, the buying decision increasingly runs through security leadership — and Anthropic is writing directly to that audience about how to say yes responsibly.
Sonnet 5 Is Now the Default in Claude Code — With 1M Context and Promo Pricing
Claude Sonnet 5 is now the default model in Claude Code, arriving with a native 1M-token context window that lets it hold far larger codebases and session histories in view. Anthropic is sweetening the switch with promotional pricing of $2 per Mtok input and $10 per Mtok output through August 31.
For developers, a default swap is a big deal: it changes the price/performance baseline for every new session without anyone opting in. The bigger context plus discounted rates make deep, whole-repo workflows meaningfully cheaper for the next six weeks.
API Rate Limits Jump; Usage Tiers Consolidated to Three
Anthropic raised rate limits across the Claude API. Sonnet and Haiku limits now match Opus at every usage tier, removing a long-standing ceiling that pushed some workloads toward the pricier model just for headroom. The tier structure has also been simplified into three levels: Start, Build, and Scale.
Alongside the throughput bump, the platform added API key expiration settings earlier in the week and an Admin API beta for Enterprise orgs to list members, change roles, and manage invites and groups. Taken together, it’s a steady push to make the platform easier to scale on and easier to govern.
Fast Mode for Opus 4.7 Is Being Removed July 24
A heads-up for anyone still on the older flagship: Fast mode for Claude Opus 4.7 has been deprecated, with removal scheduled for July 24, 2026. After that date, requests to claude-opus-4-7 carrying speed: "fast" will return an error rather than silently falling back.
If any of your production calls pin that combination, now is the window to drop the speed parameter or migrate to a current model. It’s a small change, but the kind that breaks a pipeline quietly if it slips past a release checklist.
Claude Partner Network Adds a Services Track and Partner Hub
Anthropic expanded its Claude Partner Network with two pieces aimed at enterprise buyers. A tiered Services Track ranks consulting firms by how deeply they’ve deployed Claude for clients, and a Partner Hub portal helps companies find qualified implementation partners for their rollouts.
The clever bit for the technical crowd: a new MCP connector links the Hub directly into Claude, so you can query partnership status and find implementers from inside the model itself. It’s Anthropic building the go-to-market plumbing an enterprise ecosystem needs — and doing it MCP-native.
Claude Hit With a Brief Outage on July 17
Claude and Anthropic services went down for hundreds of users on July 17, with outage reports climbing in the mid-afternoon before recovering. It was a short-lived disruption rather than a prolonged incident, but a reminder of how much daily work now routes through a single assistant.
As Claude moves deeper into classrooms, labs, and production codebases, reliability stops being a nice-to-have and becomes part of the pitch. Expect uptime to draw more scrutiny the more indispensable these agents become.
Anthropic Is Playing for the Institutions, Not Just the Chat Window
Look at today’s launches together and a pattern jumps out: teachers, scientists, CISOs, consulting partners. None of these are consumer features — they’re bids to embed Claude inside institutions with long memories and slow procurement. Claude for Teachers comes with a research pilot; Claude Science leads with auditable artifacts; the security post speaks the language of risk committees. This is what it looks like when a company stops selling a chatbot and starts selling infrastructure.
The developer news rhymes with it. Making Sonnet 5 the Code default, lifting Sonnet and Haiku rate limits to Opus levels, and simplifying tiers to Start/Build/Scale all lower the friction of standardizing on Claude at scale. The through-line is durability: pilots that generate evidence, artifacts you can audit, limits that don’t force awkward model choices. Flashy model swaps win a news cycle; being the safe institutional default is how you win the decade. The one crack in the story is uptime — yesterday’s outage is a reminder that indispensability cuts both ways.