Lead Analysis
Anthropic ships Fable 5.2 and Mythos 5.2 at $8/40 as its IPO roadshow opens
Tuesday, November 10, 2026: Anthropic launched Claude Fable 5.2 and Claude Mythos 5.2 on Monday at $8 per million input tokens and $40 per million output -- 20% cheaper than the prior generation and 20% below OpenAI's GPT-6 Astra ($10/$50) -- as its IPO roadshow formally began today.
Both models posted record benchmark scores: FrontierSWE v2 69.9% (highest among evaluated models), OSWorld 2.1 89.3%, Humanity's Last Exam with tools 74.2%, and the formalization of Fermat's Last Theorem in just 30 hours. Output speeds up to 30% faster than 5.1; optional fast mode at 2.5x speed for $16/$80. Batch API prices drop to $4/$20 -- same as standard interactive Opus 5.5 pricing.
Fable 5.2 was released generally available while Mythos 5.2 remains gated to vetted cybersecurity and life-sciences organizations through expanded verification programs -- the latter having more relaxed guardrails. Anthropic reports the models completed otherwise-infeasible software migrations, designed three working electronic devices from scratch, solved two open problems in black-hole physics, made substantial progress on robotics evaluations, formalized Fermat's Last Theorem in 30 hours, and wrote more than 90% of changes merged into Anthropic's own core code. One model can command 1,000 agents, the company claims.
The timing is tightly aligned with Anthropic's IPO. The quiet period ended November 9, Bloomberg reports, with formal marketing kicking off as early as Tuesday -- today. A reported JPMorgan investor presentation is scheduled for November 13; Nasdaq debut targeted as early as the week of November 16 or next Thursday/Friday (Nov 19-20). Pre-IPO shares have climbed past post-election levels, CryptoBriefing reports. Morningstar says the company could be worth $3 trillion (US) under its most bullish scenario (Australian Financial Review, Nov 10). The amended S-1/A disclosed $16.3 billion first-half revenue.
The pricing shift is material for Indian enterprise procurement: at $8/$40, Fable 5.2 undercuts GPT-6 Astra by 20% on both axes while outperforming it on most evaluated capabilities (FrontierCode 2.0 49.1% vs 42.8%; FrontierSWE v2 69.9% vs 65.5%). The batch rate of $4/$20 equals Opus interactive, making long-context and agentic workloads cost-effective via asynchronous routes. Combined with CERT-In's recent acquisition of Anthropic Mythos via Amazon Bedrock in India and MeitY official disclosures about agent escapes, Indian enterprises should add Anthropic's IPO calendar, pricing stability, and contract-duration due diligence to their AI-procurement checklist alongside vendor financial strength -- following OpenAI's $1.25T markdown.
Markets: Indian indices closed the first full Samvat 2083 session at eight-month highs on Monday (Sensex +107.25 to 81,106.65; Nifty +58.25 to 25,044.65). PSU banks led (+3.73%), IT slid -1.2%. The rupee broke below 94 to a six-month high near 93.70 as crude plunged ~4.5% on de-escalation hopes. The Anthropic pricing cut and open-roadshow week provide a catalyst for the IT basket if the AI-capex read-through accelerates this week.