Day 7 receipt
Claude Opus 5.5
Anthropic's first Claude 5.5 model, pitched as Fable 5.1-level work at a lower price than Opus 5, with higher plan limits.
Launch dayWhat they claimed, quoted word for word · 5
“It performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5.”
No independent cost study found. One detailed report measured more thinking and output tokens per request from 1 Oct, which would eat into the saving.
“Input and output tokens are $4 and $20 per million, 20% less than Opus 5.”
A list price. No user report disputes it; we have not checked anyone's bill.
“In addition to the price drop, we’re increasing five-hour usage limits on Pro, Max, Team, and seat-based Enterprise plans.”
Reports describe weekly and session limits draining faster after launch. Weekly limits are not the five-hour limit the claim names, so this is not a direct contradiction.
“It is much less likely than recent models to take hard-to-reverse actions or act outside the boundaries it’s been given”
One detailed report of scope creep on a bounded task list; no broader evidence either way.
“Opus 5.5 also generates output more than 30% faster than Opus 5.”
No user speed measurements found.
Day 7What real users report · 7
A Claude Code user compared local transcripts week over week and found the weekly limit draining about 3.6x faster after the 25 Sep reset with the same workflow and model mix.
Open as of 16 Oct · 12 comments · Open the source
Another user reports weekly and session limits going much faster than the previous week while using a single conversation.
Open as of 16 Oct · 9 comments · Open the source
A user with an unchanged setup measured about 2x thinking tokens and 1.6x output per request from 1 Oct, with worse judgment, and says the same shift showed up in claude.ai chat.
Open as of 16 Oct · 6 comments · Open the source
Switching a long project to Opus 5.5 led to steady scope creep: given five bounded tasks, it kept adding side-issues instead of finishing, and the user switched back to an older model.
Open as of 16 Oct · 10 comments · Open the source
On launch day, an ordinary UI question was stopped by the new safeguards with a 'reasoning_extraction' flag. A similar false positive was filed on 16 Oct.
Open as of 16 Oct · 7 comments · Open the source
Through AWS Bedrock in Cursor, long multi-step agent turns left chats permanently failing with an invalid thinking-signature error. Cursor staff confirmed it and suggested switching the chat to Opus 5 for now.
Staff-confirmed; no fix posted as of 28 Sep · Open the source
A Cursor user welcomed the pricing on launch day; Cursor staff confirmed the model was live in the picker.
655 views · Open the source
Verified fixes (from Unstuck)1
Claude Opus 5.5 / Fable returns stop_reason 'refusal' with category reasoning_extraction
Anthropic's docs say this category fires when a prompt, skill or tool schema asks the model to write its reasoning into the output (scratchpads, 'reasoning'/'thinking' JSON fields). It has no fallback model and is billed, so change the prompt: ask for a short explanation instead, and read summarized thinking via display: "summarized".