Claude Lands In-Country Inference in India via Amazon Bedrock
Anthropic will let Indian organizations run requests to Claude on servers inside India in the coming weeks, delivered through Amazon Bedrock, AWS’s platform for building and deploying generative AI. Requests routed to the India endpoint stay on domestic infrastructure — so sensitive data no longer has to leave the country to be processed.
That’s aimed squarely at the sectors that have been slowest to move: banking, insurance, telecom, and government, where data-sovereignty and compliance rules have stalled production AI. The move brings Anthropic level with OpenAI and Google, both of which already offer local processing in India, and builds on the Bengaluru office it opened in February. Anthropic now calls India Claude’s second-largest market, and is running its first partner certification in Bengaluru this week with a goal of certifying 5,000 professionals.
Cognizant Reaches Global Premier Partner Status
Cognizant has become a Global Premier Partner — the top tier of the Claude Partner Network — deepening a relationship that started in late 2025. It’s embedding Claude across its own engineering and business platforms, including Flowsource and Neuro AI, and has now put more than 30,000 associates through Claude training under a new certified-workforce model.
The receipts are what make it notable: an agentic contract-intelligence system for a biopharma client cut review time by up to 40%, and a risk-navigation tool for underwriters saves roughly eight hours a week each. It’s a sign of how the big systems integrators are turning Claude into billable, deployed enterprise work rather than pilots.
Opus 4.1 Retires August 5 — Migrate to Opus 4.8
Tomorrow is the cutoff: Claude Opus 4.1 is being retired on August 5, and Anthropic is pointing any remaining users to Opus 4.8. If you have production traffic or scripts still pinned to the opus-4-1 model string, swap it before the endpoint stops answering.
It’s part of a broader tidy-up on the platform. The legacy Workbench and the experimental prompt-tools APIs both sunset August 17, so this is a good week to audit anything hardcoded to older surfaces. The upside of migrating: Opus 4.8 is where the newer capabilities — like mid-conversation system messages — are landing.
Claude Code v2.1.221 Adds a Focus View and Sandbox Masking
Fresh today: Claude Code v2.1.221 adds a Focus view chat-menu toggle that hides tool activity behind an expandable per-turn summary — with a live running-tool indicator so you still know work is happening. It’s a cleaner read on long agentic sessions where the tool chatter can bury the actual conversation.
The release also adds a “mask” mode for sandbox credential files on Linux and WSL, tightening how secrets are handled inside the sandbox. On the API side, the advisor tool now takes a max_tokens parameter to cap the advisor model’s output per call — a small lever for trimming latency and token cost.
Free Claude Max for Open-Source Maintainers
Anthropic’s Claude for Open Source program is giving qualifying maintainers and core contributors six months of Claude Max 20x — the $200/month top tier — free, a $1,200 value, with up to 10,000 spots. Eligibility runs to primary maintainers of public repos with 5,000+ GitHub stars or 1M+ monthly NPM downloads, plus a real carve-out for low-visibility but high-dependency critical infrastructure.
One caveat worth flagging: it’s a personal-use plan, not API access — automated pipelines still bill separately. But for the humans who keep the dependency tree standing, it’s a meaningful subsidy, and a smart bit of goodwill toward the ecosystem Anthropic’s own tooling runs on.
AI for Science: Rare-Disease Grant Window Just Closed
Anthropic’s AI for Science rare-disease grants closed to applications on August 2. The program offers up to $50,000 in Claude credits over six months across two tracks — one for basic researchers, one for early-stage biotechs speeding clinical development for rare genetic conditions.
The existing grantees hint at the ambition: Every Cure is mining millions of drug-repurposing candidates, the Centre for Population Genomics is drafting variant classifications for expert review, and the Violet Research Institute is using Claude to navigate FDA guidance and draft regulatory filings. If you missed the deadline, it’s worth watching for the next cohort.
The Growth Story Is Now About Geography and Governance
Look past the model numbers and today’s news is about where Claude runs, not how smart it is. In-country inference in India, a Global Premier integrator in Cognizant, a partner-certification push in Bengaluru — these are all moves to make Claude deployable inside the constraints that regulated enterprises actually operate under. Sovereignty, compliance, and a trained delivery workforce are the unglamorous unlocks that turn interest into signed production contracts.
It rhymes with the enterprise plumbing story we’ve tracked all week — governed data, managed auth, deployable connectors. The frontier labs have largely stopped competing on benchmarks alone; the fight has moved to distribution and trust. India is the clearest tell: matching OpenAI and Google on local processing isn’t a capability play, it’s a market-access one. Whoever removes the most friction for the buyer — not whoever posts the highest eval — wins the enterprise. That’s the phase Anthropic is playing now.