AI News — June 30, 2026

Model releases

• [2026-06-29] Anthropic / Microsoft — Claude is now generally available in Microsoft Foundry on Azure. Claude Opus 4.8 and Claude Haiku 4.5 ship in the Messages API with core capabilities (prompt caching, extended thinking), running on NVIDIA GB300 Blackwell Ultra GPUs, and inference can be pinned to a US data zone for data-residency requirements — Claude runs inside the customer’s Azure environment using the auth, billing, and governance controls they already have. It matters because it gives Azure-native enterprises first-party Claude access without leaving Microsoft’s governance plane — and the model lineup is telling: with Fable 5 and Mythos 5 still export-restricted (day 18 of the blackout), Foundry launches with the tier Anthropic can actually ship, Opus 4.8 and Haiku 4.5, rather than the frontier models. (official, Microsoft Azure)

  Generally available is the line that matters more than the feature list: GA is when Claude on Foundry carries production support and can be folded into an existing Azure enterprise agreement, so for Azure-committed shops this is the point it shifts from a model you can pilot to one procurement can actually sanction.

  For Solution Architects: A regulated workload that must keep inference in-region can now pin Opus 4.8 to the US data zone inside the customer’s own Azure environment, under the same auth, billing, and governance controls the rest of the stack already uses — so adding Claude stops meaning a separate Anthropic or Bedrock integration with its own data-residency and governance review.

Coding agents

• [2026-06-29] Cursor — Cursor Mobile opened as a public iOS beta on all paid plans. From the phone you can pick a repo and launch always-on agents in the cloud or drive an agent running on your computer, then get notified when work is ready to review — the same agent-launch flow as the desktop app, built for supervising long-running work away from the keyboard. It lands alongside the broader push from Cursor’s Compile conference: Origin, a Git platform pitched as a GitHub competitor “built for the agentic era,” and a first in-house model trained from scratch (reportedly with ~10× more compute) that Cursor says is weeks from shipping. It matters because agent oversight is moving off the desktop — the bottleneck Cursor is betting on is review and unblocking, not editing. (official, TechCrunch)

  Driving an agent on your own computer — not just the cloud agents — is the more interesting half: your local environment, with its secrets, configs, and working-tree state, stays the execution context while the phone acts as a remote control, so mobile oversight doesn’t require reproducing your toolchain in a cloud sandbox first.

  For Software Developers: A build goes red while you’re away from your desk — open the iOS beta, pick the repo, and launch a cloud agent to dig into the failure right then, with a notification when its fix is ready to review, rather than letting the broken build sit untouched until you’re back at the keyboard.

Watch list

• Fable 5 general release — day 18 of the blackout and the consumer-grade model is still dark. Axios’s June 27 reporting put restoration “as soon as this coming week” with Anthropic confident in “the coming days,” but the Pentagon and NSA still have to clear it, and the July 8 consumer-terms update enumerating identity-verification data remains the most-cited mechanism for a US-users-first return. Watch for an Anthropic announcement and whether access returns gated by verified identity or as a plain public unlock. (Axios, prior coverage)

  Axios filed the “as soon as this coming week” line on June 27, so the window it pointed at is the one opening now — which makes the next few days the test of whether the reporting holds or the date slips, as the GA timelines for every other gated model on this list already have.

• Gemini 3.5 Pro (GA) — month-end arrives today with the model still a limited Vertex AI enterprise preview: no model card, no public benchmarks, and no pricing for the promised 2M-token context and Deep Think mode. Pichai’s Google I/O “give us until next month” has now lapsed, and the slip comes amid a run of senior DeepMind departures (John Jumper, Jonas Adler, and Alexander Pritzel reportedly bound for Anthropic; Noam Shazeer for OpenAI) that sent Alphabet shares down ~5% on June 22. Watch for whether a rate card and model card land in July or the GA target slides further into Q3. (Fortune, prior coverage)

  With today being the month-end that Pichai’s own “give us until next month” pointed to, a July rate card would already be a second missed self-imposed date — and paired with the senior-DeepMind departures the bullet notes, the slip starts to read as a capability-or-staffing question rather than final polish.

• GPT-5.6 general availability — Sol, Terra, and Luna remain a ~20-partner government-coordinated limited preview; OpenAI says GA is “in the coming weeks.” Watch for the GA date, whether the vetted-partner gate persists at GA, and published third-party benchmarks beyond the internal CTF cyber scores. (prior coverage)

• Alibaba / Qwen distillation amendment — the Hagerty–Kim amendment to must-pass defense legislation (which would blacklist or sanction Chinese firms found improperly accessing US AI-model output) was floated for as early as June 24; that date has passed without a confirmed filing. Watch for whether it has been formally entered into the bill’s text and Alibaba’s still-absent response. (prior coverage)

• MCP spec finalization (July 28) — the 2026-07-28 release candidate (stateless core, Extensions framework, Tasks, MCP Apps, authorization hardening, formal deprecation policy) is in its validation window, with the Ruby/TypeScript/Python SDKs being updated against it. Watch for the rest of the SDK support landing — and any breaking issues surfaced in Tasks, MCP Apps, or the new MCP-specific HTTP headers — before the cutover. (official, prior coverage)