AI Daily · Nov 10, 2026: Anthropic releases Claude Fable 5.2 with 20% lower pricing
Anthropic launched Claude Fable 5.2 at $8/$40 per million tokens, 20% below 5.1, with 1M context and 128K output; Meta also shipped its small Muse Ember model at just $0.10/$0.50 per million tokens.
Pick a date
-
Anthropic releases Claude Fable 5.2 and Mythos 5.2, 20% cheaper
Claude Fable 5.2 is available on all paid plans, the Claude Platform and major clouds, with always-on adaptive thinking, a 1M-token context window, up to 128K output and up to 30% faster generation than 5.1; it is priced at $8 per million input tokens and $40 per million output, down from $10/$50. Claude Mythos 5.2 shares the same underlying model but is limited to cybersecurity or biomedical orga
-
Meta releases Muse Ember, a small model for agent workloads
Muse Ember is Meta's smallest and most efficient model, built for high-volume agentic work such as tool calls, coding subtasks, routing, extraction and document parsing. It has a 1M-token context window, natively handles text, images, audio and video, and is available in Muse Code and on the Meta Model API. The earlier stealth model Crouton Alpha was an early version of it; over 5 trillion tokens
-
Muse Ember priced at $0.10/$0.50 per million tokens
Meta set Muse Ember API pricing at $0.10 per million input tokens and $0.50 per million output tokens, available from launch day.
-
Fable 5.2 cache reads at $0.20/M; Cursor warns effort levels behave differently
Fable 5.2 is priced at $8 per million input tokens and $40 per million output tokens, 20% less than Fable 5.1, with cache reads at $0.20 per million, a 1M context window and 128K output. Requesty reports it runs up to 30% faster than Fable 5.1, and Cursor notes effort levels behave differently, so users should start from the default setting.
-
Fable 5.2 hits 69.9% on FrontierSWE v2, 30% faster output
On OpenRouter, Claude Fable 5.2 offers 1M context at $8/M input and $40/M output, with up to 30% faster output than Fable 5.1; Anthropic reports 69.9% on FrontierSWE v2, 89.3% on OSWorld 2.1 and 74.2% on Humanity's Last Exam.
-
Vals Index shows Ember cheapest at $0.15 per test, fastest in top 12
On the Vals Index, Meta's Ember costs $0.15 per test, the lowest of any model, beating Haiku 5.5 and GPT-6 Luna, and finishes tests in about 15 minutes, the fastest among the top 12.
-
Muse Ember goes live in Command Code and on Google Cloud with 1M context
Muse Ember is now available in Command Code on all plans, with a 1M token context window and pricing at GPT-6 Luna and Haiku 5.5 levels; the $10/mo GOAT plan includes $20 of usage. It also landed on Google Cloud alongside Muse Wonder and Muse Spark 1.3. Ahmad Awais says Muse Ember is 17x cheaper than Muse Spark 1.3 (max) at near-similar intelligence.
-
Fable 5.2 arrives in Perplexity Computer and Search
Claude Fable 5.2 is now in Perplexity Computer for Pro and Max subscribers and in Search for Max subscribers, scoring a record 0.851 on Perplexity's WANDR benchmark at $14.51 per task.
-
Fable 5.2 tops CursorBench at 60% on max effort
Claude Fable 5.2 is available in Cursor and scores 60% on CursorBench at max effort; at default effort it reaches 56.9%, up from 49.2% for Fable 5.1.
-
Meituan LongCat 2.5 free on OpenCode for two weeks
Meituan's LongCat 2.5 is available free on OpenCode for two weeks, offering 1M context, multi-modal input and zero data retention.