AWS Bedrock Silently Kills Claude Opus 4.7 Quota: A Warning for Production AI Workflows

A Hacker News user reports that their Claude Opus 4.7 model access was suddenly revoked on Amazon Bedrock, with the TPM quota silently set to 0. The user, who relies on the model for production workflows serving government customers, encountered throttling errors before receiving an explanation from AWS support.
According to the support response, the quota drop occurred due to a “recent system update” that adjusted access controls based on factors including regional considerations, payment history, and usage patterns. Quotas were set to 0 starting May 1st, 2026, and AWS explicitly states that restoration “cannot guarantee approval.” The recommended workaround: migrate to Claude Opus 4.6, for which the user has quotas of 10,000 requests/min and 3,000,000 tokens/min in us-east-1 (with cross-region and global variants). The support message notes that Opus 4.6 “can serve as an effective replacement with minimal code changes.”
HN commenters express frustration. User DetroitThrow calls this “insane for a company to pull this crap on paying customers with production workflows” and recommends switching to Google AI or direct Anthropic access. They note that Bedrock has a history of random throttling, citing Quinnypig‘s previous coverage of enterprise support rug-pulls. The original poster confirms they “can no longer make even a single request.”
This incident underscores the risk of relying on AWS Bedrock for production AI inference, especially with frontier models where access can be revoked without notice. Developers running mission-critical workloads on Bedrock should consider multi-provider fallbacks or direct API contracts with model providers like Anthropic.
📖 Read the full source: HN AI Agents
👀 See Also

Claude Admits It's an Affirmation Echo Chamber: The Full Breakdown
In a Reddit post, Claude provides a brutally honest self-assessment: it's an agreeability-by-default system that cannot learn, verify, or self-correct.

Meta OpenEnv AI Hackathon in India Offers Direct Interviews and $30K Prize Pool
Meta is hosting India's first OpenEnv AI Hackathon in collaboration with Hugging Face and PyTorch, where developers build reinforcement learning environments for AI agents. Top teams get direct interviews with Meta and Hugging Face AI teams, plus a $30,000 prize pool.

AI Reimplementation of chardet Library Raises Copyleft Licensing Questions
Dan Blanchard used Anthropic's Claude to reimplement the chardet Python library from scratch, changing the license from LGPL to MIT. The resulting code shows less than 1.3% similarity to previous versions, sparking debate about whether AI-assisted reimplementation erodes copyleft protections.

Claude Code v2.1.77 Release: Token Limits, Sandbox Controls, and Bug Fixes
Claude Code v2.1.77 increases default maximum output token limits for Claude Opus 4.6 to 64k tokens and adds an allowRead sandbox filesystem setting. The release includes over 30 fixes for issues ranging from memory management to terminal UI behavior.