
AI News / Claude · Nov 2026
Claude Fable 5.2 and 4 more November updates
Anthropic released Fable 5.2 on November 9 for demanding technical work. Here is when it makes sense to try it.
↓ Swipe up · 5 facts

1MToken context window
Fable 5.2 is for demanding, multi-step work.
Anthropic released it on November 9. It shares an underlying model with Mythos 5.2, which requires verified access. Fable is generally available with additional safeguards for cybersecurity and biology. The API supports a 1 million token context window and up to 128,000 output tokens.
Source: Anthropic ↗
Read-onlyNo money can be moved
Claude now reads your spending.
Launched on November 5 for US subscribers in beta, Personal Finance connects bank and credit card accounts through Plaid. Access is read-only: Claude can analyse spending but can't move, manage or invest money.
Source: TechCrunch ↗
IPONasdaq filing
The company behind Claude is going public.
Anthropic publicly filed for an initial public offering on Nasdaq. Coverage has focused on the S-1, including questions from a US senator about how existential-risk disclosures are worded.
Source: Bloomberg ↗
25+Partner agents & products
25+ partner agents, one marketplace.
Anthropic expanded the Claude Marketplace with more than 25 partner agents and products across security, development, creative work and productivity, alongside new enterprise security and admin controls.
Source: Releasebot ↗
BetaThreads · IP allowlisting
Threads, allowlists and watermarks for builders.
The Claude Developer Platform added beta message threads in the Messages API and beta IP allowlisting in the Console, plus watermarking of generated text for several models. Fast mode on an older Opus model is being retired on November 30.
Source: Releasebot ↗
Arrow's take
What it means for brands
- Start with Opus 5.5 for most coding or knowledge work. Try Fable 5.2 on a difficult migration, a long research task or a technical analysis when a simpler run falls short.
- Compare the two on the same inputs, with the same access and acceptance criteria. Score evidence coverage, corrections, time and cost before changing your default.
- For API use, Fable 5.2 costs $8 per million input tokens and $40 per million output tokens at standard rates. Opus 5.5 costs $4 and $20 respectively. Check your data-retention requirements before running sensitive work.
60-second story · AI News / Claude
Claude Fable 5.2 is Anthropic’s model for demanding reasoning and long-running technical work. Anthropic recommends starting with Opus 5.5 for most coding and knowledge tasks, then testing Fable on the hardest cases.
November 2026 edition
Perplexity Teammates: 5 facts about the AI agents that never log off