Anthropic released Claude Opus 4.7 on April 16 — the newest flagship model, available across the API, Amazon Bedrock, Google Vertex AI, and Microsoft Foundry. The official improvements are real: stronger multi-step task performance, more reliable agentic execution, improved long-horizon reasoning, and the first Claude model with high-resolution image support (max resolution up to 2576px / 3.75MP from 1568px / 1.15MP). A new "task budgets" feature gives Claude a token-level estimate for agentic loops. Pricing holds at $5/M input, $25/M output — though a new tokenizer may use up to 35% more tokens on some content, making effective costs modestly higher. The claims on design, presentation, and visual understanding look strong. However, early user reports are less uniformly positive: multiple users and reviewers have flagged that Opus 4.7 appears to hallucinate more frequently than Opus 4.6, and some have reported that overall task performance feels like a regression in certain workflows. This is worth watching — model releases often improve on benchmarks while introducing new failure modes in production use. As with any frontier model update, the recommendation is to test against your own use cases before assuming the upgrade is strictly better.
Claude Opus 4.7 Ships — Impressive Design, But Early Users Report More Hallucinations
Anthropic / GitHub / FindSkill.ai
Industry
More on AI News
- The Week's Two Loudest Headlines: Huang Out at Nvidia, and the Outlook Outage November 16, 2026
- Anthropic's IPO Math: A $17B Quarter Under $895.4B of Compute Commitments November 16, 2026
- Agents Get a Wallet November 16, 2026
- AI Infrastructure Economics: A Power-Gated Supply Meets Compounding Demand July 5, 2026
Get the newsletter in your inbox
A weekly, curated digest on AI and organizational design.