Commerce Dept lifts export controls on Claude Fable 5 and Mythos 5 effective July 1 2026
US government reversed export controls on two Anthropic frontier models, enabling global API access from July 1. The move deviates from the prior tightening trajectory and resets compliance baselines for developers and cloud providers. It signals a recalibration of controls that may extend to additional models within months.
Anthropic received formal notice from the Department of Commerce that export controls on Claude Fable 5 and Mythos 5 were rescinded. The company stated restoration of access begins July 1. The decision reverses prior restrictions placed on frontier-scale models in 2025 under updated EAR rules targeting training compute above 10^26 FLOPs.
Primary evidence is the Anthropic notification citing direct Commerce correspondence. No public BIS Federal Register notice has appeared yet. Related actions include simultaneous de-listing of select inference hardware SKUs previously tied to the same control list entries.
This marks the first documented reversal of a frontier-model export restriction since the 2023-2025 expansion of CCL Category 4. Operational impact includes immediate re-enablement of API endpoints for non-US users previously blocked by geofencing. Downstream, it reduces friction for international fine-tuning workloads that had shifted to domestic alternatives.
Next steps center on whether BIS issues clarifying guidance on compute thresholds or applies the same treatment to competing models from OpenAI and Google. Regulated entities are monitoring for reciprocal actions from Wassenaar Arrangement partners.
BIS: Additional frontier model controls lifted on one OpenAI system by September 2026
Sources (3)
- [1]Primary Source(https://twitter.com/AnthropicAI/status/2072106151890809341)
- [2]Supporting Source(https://www.bis.doc.gov/index.php/documents/federal-register-notices/2026)
- [3]Supporting Source(https://www.ecfr.gov/current/title-15/subtitle-B/chapter-VII/subchapter-C/part-744)