Claude Fable 5 Is Back: Export Controls Lifted, API Live Again July 1

Anthropic said on June 30, 2026 that the US Department of Commerce has lifted its export controls on Claude Fable 5 and Claude Mythos 5, ending a 19-day blackout that pulled two frontier models offline mid-flight. Fable 5 came back globally on July 1. (Source: Anthropic official X post, 2026-06-30)

Key facts:

• The blackout ran 19 days. Anthropic disabled Fable 5 and Mythos 5 on June 12 to comply with the export-control directive, and restored access July 1. (Source: Al Jazeera, 2026-07-01)
• Fable 5 is available globally again. It returned across the Claude Platform, Claude.ai, Claude Code, and Claude Cowork on July 1. (Source: CNBC, 2026-06-30)
• There is a usage cap until July 7. Pro, Max, Team, and select Enterprise users get Fable 5 at 50% of existing usage limits through July 7; after that the model moves to usage-based credits. (Source: Al Jazeera, 2026-07-01)
• The safeguards were strengthened first. Anthropic worked with the US government and Amazon to harden Fable 5’s cyber protections before flipping it back on. (Source: Forbes, 2026-07-01)
• The trigger was a jailbreak report. An Amazon research report showed a prompt could bypass some of Fable 5’s safeguards and surface software-vulnerability information — that is what prompted the June 12 controls.

What comes back online: Fable 5’s agentic-coding scores — SWE-Bench Pro 80.3% and FrontierCode 29.3%, both ahead of Opus 4.8. (Source: Anthropic, 2026-06-09)

Why this took a government letter

Fable 5 ships as the same weights as Mythos 5 with safety classifiers that route flagged cyber, bio, and chemistry requests down to the weaker Opus 4.8. On Anthropic’s own offensive-cyber evaluations, Fable 5 scores 0.0 across the board — the safeguards zero it out — while Mythos 5 runs high. The government’s concern was that a narrow jailbreak could defeat that routing and reach Mythos 5’s unrestricted vulnerability-discovery behavior. The fix Anthropic negotiated was hardened safeguards plus government sign-off, not a capability rollback.

The safeguard that resolved the standoff: Fable 5 (orange) reads 0.0 on every offensive-cyber test; Mythos 5 (pink) is the unrestricted model. (Source: Anthropic, 2026-06-09)

What this means if you’re building with Claude Fable 5

If you pinned claude-fable-5 in a workflow before June 12 and swapped to a fallback when it went dark, the model id resolves again as of July 1 — but read the rollout before you flip production back:

• Mind the 50% cap until July 7. If you re-enable Fable 5 across a Pro/Team seat this week, you are working against half your normal usage limit. Heavy CI or agent loops can burn it fast and start erroring mid-run.
• The pricing model shifts July 7. After the cap lifts, Fable 5 moves to usage-based credits — budget and alerting you set during the free/limited window won’t match what you pay next week.
• Keep the fallback chain in config anyway. This is the second forced Fable 5 availability change in three weeks. A model id that’s live today can be pulled by government order again; leave claude-opus-4-8 wired as the automatic fallback. Our guide to switching to Opus 4.8 and the Claude Code daily-driver setup cover the swap either direction.

For the full arc, see our coverage of the Fable 5 launch, the June 12 export-control suspension, and the June 26 partial re-release to ~100 US firms — this July 1 lift is what finally reopened Fable 5 to everyone else.

Sources

• Anthropic — “Commerce has lifted export controls on Claude Fable 5 and Mythos 5” (official X post, 2026-06-30)
• US lifts restrictions on Anthropic’s powerful AI models Fable and Mythos — Al Jazeera, 2026-07-01
• Anthropic says Trump admin has lifted export controls on Claude Fable 5 and Mythos 5 — CNBC, 2026-06-30
• US Lifts Restrictions On Anthropic’s Mythos 5 And Fable 5 AI Models — Forbes, 2026-07-01