← Back to feed

Anthropic extended Claude Fable 5 free access on all paid plans through July 19, 2026, announcing the reprieve via its official @claudeai account on the night of July 12 — hours before the previous deadline was set to expire. The 50 percent weekly usage cap remains in place.

The extension is the third in five weeks. Fable 5 launched June 9 with free access promised through June 22. An export-control directive shut both Fable 5 and Mythos 5 down globally on June 12. After the Commerce Department cleared both models on June 30, Anthropic restarted the free window July 1, capped at 50 percent of weekly usage. On July 7 — hours before that deadline — the first extension appeared, running through July 12. On July 12, hours before expiry, the third extension arrived: July 19.

Claude Code lead engineer Thariq has stated publicly that Anthropic aims to return Fable 5 to standard subscriptions when compute capacity allows. No timeline has been attached to that commitment and the capacity constraint has not resolved in five weeks.

The Honeycomb Leak

On July 8, developer @chetaslua posted screenshots of a model called “Claude Honeycomb EAP” that had materialized inside Cursor’s model selection menu and then disappeared within hours. Two prompts ran before the model was pulled.

The briefly glimpsed spec sheet described it as an “Anthropic research model with per-turn controls and safety fallbacks,” available as an Early Access Preview, with a one-million-token context window and a version labeled “extra high effort.”

The detail that spread furthest: the model routes sensitive queries to Claude Opus 4.8 as a safety fallback rather than handling them directly. In Fable 5’s published architecture, safety classifiers for cybersecurity and biology requests trigger exactly that fallback, routing to Opus 4.8 in under five percent of sessions. A fallback chain that steps down to Opus 4.8 implies Honeycomb sits above Opus 4.8 in capability.

The spec matches Fable 5’s documented architecture on every point: one-million-token context, extra-high-effort mode, safety classifiers routing to Opus 4.8. The model string does not appear in Anthropic’s public API or documentation. EAP models appearing in Cursor have preceded public launches before. They have also been pulled without shipping.

Community consensus currently places Honeycomb as Opus 5 or a direct Fable 5 successor, with a possible end-of-July launch. Anthropic has not confirmed, denied, or commented.

The Competitive Context

The rolling extensions are not administrative. They track competitive pressure. The first extension came hours before the July 7 deadline. The second arrived on the same weekend Grok 4.5 and GPT-5.6 launched.

Two well-resourced rivals have moved into Anthropic’s Opus-tier market with models priced at roughly one-eighth to one-sixth of Fable 5’s $10/$50 per million output cost — GPT-5.6 Sol at $5/$30 and Grok 4.5 at $2/$6 — and carrying benchmark results credible enough to force the comparison.

Fable 5 retains the clearest benchmark lead on the evals closest to real repository-scale engineering. Opus 4.8 outperforms both Grok 4.5 and GPT-5.6 Sol on SWE-bench Pro. But the cost-efficiency argument has shifted structurally in July 2026. Whether Anthropic’s next move changes that calculus depends on what Honeycomb actually benchmarks.

For now, the model is two prompts and a vanished listing. The July 19 deadline is the next date that matters.