Apple Offers Free Private Cloud Compute to Indie Developers with Under 2M Downloads
Apple announced at WWDC 2026 that developers with fewer than 2 million first-time App Store downloads can use its Foundation Models running in Private Cloud Compute at no cloud API cost. The move is aimed at indie developers who experiment with AI but are held back by infrastructure costs.
Key Details
- Eligibility: under 2 million first-time App Store downloads.
- Access to Apple's Foundation Models with privacy-preserving Private Cloud Compute.
- Free tier covers cloud API costs — no additional cloud bills.
- Foundation Models framework expands this year to include image input and support for server models.
- Developers can integrate with their cloud model provider of choice for more complex tasks.
Apple positions this as a way to make getting started with AI as affordable as possible, contrasting with broader industry belt-tightening. Tech giants like Meta and Amazon have discontinued internal AI token usage leaderboards, and Uber ran through its 2026 AI budget in just four months.
Context for Developers
This is similar to Apple's Small Business Program, which offers lower commission rates to smaller developers. By waiving infrastructure fees, Apple hopes to attract indie developers to its ecosystem, offering frontier-tier intelligence with strong privacy guarantees.
Who It's For
Indie developers building AI-powered apps with fewer than 2 million installs, who want to prototype without cloud costs.
📖 Read the full source: HN AI Agents
👀 See Also
GitHub Copilot Moves to Usage-Based Billing by Token Consumption, Replacing Premium Requests on June 1, 2026
GitHub Copilot transitions from premium request units to token-based GitHub AI Credits, with plan prices unchanged. All paid plans include monthly credits equal to subscription cost; additional usage billed at API rates.
Claude Code 2.1.136: Action Safety, Hard Deny Rules, and Security Monitor
Claude Code CC 2.1.136 adds action safety and truthful reporting requirements, introduces hard_deny as a fourth custom-rule category, and splits security blocking into unconditional hard blocks and user-authorizable soft blocks.
Research shows personality affects Claude's self-correction, not Llama or Qwen
A researcher ran 23 experiments testing self-correction without guardrails across Claude, Llama, and Qwen. The main finding: personality profiles affect Claude's self-correction ability, with high directness catching all errors and low directness catching none. Llama and Qwen didn't self-correct even with identical prompts.
Claude Code developer acknowledges adaptive thinking flaw, provides workaround
Boris Charny, creator of Claude Code, confirmed a flaw in the adaptive thinking feature that causes performance degradation. Users experiencing issues even with effort=high settings can use CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING=1 as an interim workaround.