AI and security briefsModels

How much does Claude Fable 5.2 cost?

By Brandon Da Costa, founder of Digiton Dynamics

Published 9 November 2026Updated 9 November 2026Editorial policy

Claude Fable 5.2 costs $8 per million input tokens and $40 per million output tokens on the Claude API. Cache reads are $0.20, the Batch API halves both prices and fast mode doubles them. That is 20% below Fable 5.1 and twice Opus 5.5 per token.

Who should act: Teams that pay for Claude by the token and need a budget for Fable 5.2 before they move a service, a router or a Claude Code workload onto it.

What changed

Anthropic released Claude Fable 5.2 on 9 November 2026 at a lower price than Fable 5.1 (Anthropic announcement, 9 November). The rates per million tokens, in US dollars (Anthropic pricing page, 9 November):

• Fable 5.2: input $8, output $40, 5-minute cache write $10, 1-hour cache write $16, cache read $0.20.
• Fable 5.1: input $10, output $50, 5-minute cache write $12.50, 1-hour cache write $20, cache read $0.25.
• Opus 5.5: input $4, output $20, 5-minute cache write $5, 1-hour cache write $8, cache read $0.20.

Three modifiers stack on those rates. The Batch API takes 50% off input and output, so Fable 5.2 drops to $4 and $20. Fast mode costs $16 and $80, cannot be combined with batch and, by Anthropic's account, speeds up output by up to 2.5 times. Requesting US-only inference with inference_geo multiplies every category by 1.1. The full 1M context window is billed at the same rate per token whatever the prompt length (Anthropic pricing page, 9 November).

Cache reads cost 0.025 times the base input price on Fable 5.2, against 0.05 on Opus 5.5 and 0.1 on most older models. A long system prompt that is reused costs the same $0.20 per million tokens on either model.

Who it affects

• Fable 5.1 users. The same work costs 20% less per token. Fable 5.2 has its own rate limit, separate from Fable 5.1's.
• Opus 5.5 users weighing a move up. Per token it is double. Anthropic's own early testing on FrontierCode 2.0 put Fable 5.2 at high effort about level with Opus 5.5 at max effort, for roughly half the cost per task. That is a vendor number on one benchmark.
• Claude Code subscribers. Plans are not billed per token. Anthropic says Pro plans and standard Team seats get Fable for up to 50% of weekly usage limits during launch week.
• Anyone who left effort unset. The API default is high, and thinking is billed as output tokens.

What to do now

1. Price one real task. Our sum for a request with 200,000 input tokens, 150,000 of them cache reads, and 8,000 output tokens including thinking: about $0.75 on Fable 5.2, $0.94 on Fable 5.1 and $0.39 on Opus 5.5, before cache writes.
2. Log usage from every response and sum it per task, not per request.

   jq '.usage | {input_tokens, cache_read_input_tokens, cache_creation_input_tokens, output_tokens}' response.json

3. Cache the stable prefix. At $0.20 a cached read costs 2.5% of a fresh one, so a long system prompt and tool list should never be resent uncached.
4. Move offline work, such as evals, classification and backfills, to the Batch API at $4 and $20.
5. Set effort per route. Start at medium for routine traffic and keep high or xhigh for the tasks that measure better there.

What to watch

• Output share. Thinking counts as output at $40, so an agent that reasons at length costs more than its prompt size suggests.
• Subagents. Anthropic warns that delegation can account for a large part of what a session costs.
• The launch promotion on Claude plans, which Anthropic describes as this week only.

Questions people ask

Digiton's production review checks which model, effort and caching setup each service runs and what it costs per finished task, and the checks are on the vibe coded app audit page.

How this brief was made: researched from the sources below, drafted with AI assistance and checked against each source on the dates shown. See the editorial policy.

Sources

1. Anthropic pricing page, read 9 November 2026. https://platform.claude.com/docs/en/about-claude/pricing
2. Claude Fable 5.2 model page, 9 November 2026, read 9 November 2026. https://platform.claude.com/docs/en/models/fable-5-2/overview
3. What's new in Claude Fable 5.2, 9 November 2026, read 9 November 2026. https://platform.claude.com/docs/en/models/fable-5-2/whats-new-fable-5-2
4. Anthropic announcement of Claude Fable 5.2 and Claude Mythos 5.2, 9 November 2026, read 9 November 2026. https://www.anthropic.com/claude-fable-and-mythos-5-2

More briefs

How these briefs are sourced and corrected. All AI and security briefs