Fable 5 is back. Sonnet 5 is now the default. Here is what changes for builders.
Fable 5 Returns
Reopens to everyone on July 1, across Claude Platform, Claude.ai, Claude Code, and Cowork.
- Through July 7, usage counts toward your weekly quota, capped at 50%, then switches to usage credits.
- Anthropic tightened its safety classifier, so expect more false positives on debugging and security requests.
- Blocked requests now fall back to Opus 4.8.
Sonnet 5 Is the New Default
Live in Claude Code and the API, and now the default for Free and Pro.
- Launch pricing through August 31: $2/$10 per million tokens, rising to $3/$15 after.
- New tokenizer means the same input can cost 1.0-1.35x more tokens than Sonnet 4.6.
The Catch
Theo tested it anyway. In the Artificial Analysis Intelligence Index, Sonnet 5 generated close to twice the output tokens of Opus 4.8 on the same tasks. Lower sticker price, higher real bill: its actual benchmark cost beat Opus 4.8, and in a later run, beat Fable too.
Check total token output before you assume Sonnet 5 saves you money.
Caveat: safety numbers are Anthropic's own testing; the cost comparison is one third-party benchmark, not a universal ranking.
Sources:
- Redeploying Fable 5: https://www.anthropic.com/news/redeploying-fable-5
- Claude Sonnet 5: https://www.anthropic.com/news/claude-sonnet-5
Fable 5 returns globally on July 1 US time.
My read: GPT-5.6 could go wide in the same window.
If that happens, the price OpenAI sets for GPT-5.6 against Fable 5 will matter more than which model wins the benchmark.