Claude Fable 5.2 is Anthropic's new top model, released on November 9, 2026. On the API it costs $8 per million input tokens and $40 per million output tokens, 20% less than Fable 5.1. Anthropic's launch table puts it a step above Opus 5.5 on almost every benchmark. Opus 5.5 still costs half as much per fresh token.
In October, when a Fable 5.5 entry leaked in Azure data, I posted that I did not want a new Fable yet. Opus 5.5 was enough for me, and I still run my daily work on it. Anthropic shipped a stronger Fable anyway, and made it cheaper. Here is what changed, who should move, and how I would test it before paying the higher rate.
- Price: $8 and $40 per million tokens, down from $10 and $50 on Fable 5.1. Batch costs $4 and $20.
- Speed: up to 30% faster output than Fable 5.1, plus an optional fast mode at $16 and $80.
- Plans: included up to half the weekly limit on Max, Ultra and Team premium seats. Pro and Team standard seats get the same until November 16.
- Mythos 5.2 is the same model without the cyber and biology blocks, for verified organizations only.
- My default stays Opus 5.5. Fable 5.2 is for the jobs where Opus falls short.
in
“Anthropic's Fable 5.5 leaked. I don't want it.”
What Is Claude Fable 5.2?
Fable 5.2 is the upgrade of Anthropic's premium model line, built for long, hard technical work. Claude Fable and Claude Mythos are the same underlying model. Fable carries strong cybersecurity and biology safeguards, so anyone can use it. Mythos drops those blocks and goes only to organizations accepted into Anthropic's Cyber and Life Sciences Verification Programs.
The API model ID is claude-fable-5-2. It has a 1M-token context window, 128,000 output tokens and a June 2026 knowledge cutoff, like Opus 5.5. In the API its default effort is high. Opus 5.5 defaults to medium. In the Claude apps and Claude Code, Fable 5.2 runs at medium by default.
Anthropic's own advice is worth reading twice. It says to start with Opus 5.5 as the daily driver. Move up to Fable only when the work needs it. Fable 5.2 can also sit behind smaller models as an advisor, or run a team of them as an orchestrator.
The test that shows what "long work" means: an Anthropic engineer asked Fable 5.2 to build an operating system from an empty folder. It coordinated more than 1,000 subagents for 8 days and delivered a system that boots on a laptop, with drivers, file storage, windows and an installer.
How the effort setting changes cost and quality is in my guide to Claude effort levels.
How Much Does Claude Fable 5.2 Cost?
Claude Fable 5.2 costs $8 per million input tokens and $40 per million output tokens on the Claude API. Every price line is 20% below Fable 5.1, and Mythos 5.2 costs the same. Opus 5.5 stays at $4 and $20, so Fable 5.2 now costs exactly twice Opus on fresh input and output.
| Price line | Fable 5.1 (before) | Fable 5.2 (now) | Opus 5.5 |
|---|---|---|---|
| Input | $10 | $8 | $4 |
| Output | $50 | $40 | $20 |
| Cache read | $0.25 | $0.20 | $0.20 |
| 5-minute cache write | $12.50 | $10 | $5 |
| 1-hour cache write | $20 | $16 | $8 |
| Batch input and output | $5 and $25 | $4 and $20 | $2 and $10 |
| Fast mode input and output | Not offered | $16 and $80 | $8 and $40 |
The cache line is the one I would look at first. A cache read on Fable 5.2 costs $0.20 per million tokens, the same as Opus 5.5. Agents reread the same context on every step, so on long cached sessions the real gap between the two models is much smaller than 2 times.
Here is my worked example, an outbound research agent with 3,000,000 cached tokens read, 150,000 written to a 5-minute cache, 50,000 fresh input and 10,000 output. One session costs $2.90 on Fable 5.2, $3.63 on Fable 5.1 and $1.75 on Opus 5.5. These are list prices on my assumptions. I did not measure them on a real bill.
That gives me a simple break-even rule. In this session Fable 5.2 costs 1.66 times Opus 5.5. So Fable is cheaper only if it finishes the job with about 40% fewer tokens or steps. On fresh, uncached work the bar is 50% fewer. If it can't clear that bar on your jobs, pay for it only when Opus fails.
Run your own numbers in the free Claude API cost calculator, which now includes Fable 5.2.
2 more costs to know. US-only inference adds 10% to every line. Fast mode generates tokens up to 2.5 times faster at double the price, on Claude Code, the Claude Platform, Google Cloud and Microsoft Foundry. Fable 5.2 and Mythos 5.2 also require 30-day data retention for safety monitoring, unless Anthropic authorizes zero data retention, for example through its Enterprise Frontier Safeguards beta.
Is Fable 5.2 Better Than Opus 5.5?
Yes, on Anthropic's published numbers. In its launch table Fable 5.2 beats Opus 5.5 on every row except 1 Elo score on knowledge work. There Opus leads 1800 to 1785. The biggest gaps over Opus are on robotics and remote labor tasks, where Fable 5.2 scores about twice Opus on the Remote Labor Index.
| Benchmark | What it tests | Fable 5.2 | Fable 5.1 | Opus 5.5 |
|---|---|---|---|---|
| FrontierCode 2.0 | Agentic coding | 49.1% | 32.4% | 45.5% |
| FrontierSWE v2 | Agentic coding | 69.9% | 56.3% | 62.3% |
| AutomationBench | Business workflows | 51.4% | 31.4% | 42.5% |
| Remote Labor Index | Knowledge work | 41.3% | 17.9% | 21.3% |
| Humanity's Last Exam, with tools | Hard reasoning | 74.2% | 65.6% | 67.7% |
| OSWorld 2.1 | Computer use | 89.3% | 83.5% | 87.2% |
| Chartography | Reading charts | 73.4% | 44.8% | 64.4% |
Read these with 2 warnings. They are the vendor's own tests, at each model's highest available effort. And on AutomationBench, Google's Gemini 4 Carbon scores 53.9%, above Fable 5.2. No model wins everything.
The number I care about more is cost per task. Anthropic says Fable 5.2 at high effort scored about level with Opus 5.5 at max effort on FrontierCode 2.0, for roughly half the cost per task. If that holds on your jobs, the more expensive model can be the cheaper one.
For the older comparison and the benchmarks that moved, read Fable 5.1 vs Opus 5.5.
Can You Use Fable 5.2 on Claude Pro, Max or Team?
Fable 5.2 works like the earlier Fable models on your plan. On Max, Ultra and premium Team or legacy Enterprise seats it is part of the plan, up to 50% of the weekly limit. On Pro and Team standard seats it runs on usage credits from the first message. A launch offer changes that for 1 week.
| Plan | Normal access | Launch offer |
|---|---|---|
| Max and Ultra | Up to 50% of the weekly limit | 1 to 3 Fable resets, expire March 8, 2027 |
| Team and legacy Enterprise premium seats | Up to 50% of the weekly limit | 1 to 3 Fable resets, expire March 8, 2027 |
| Pro | Usage credits only | Up to 50% of the weekly limit until November 16, plus 1 reset |
| Team standard seat | Usage credits only | Up to 50% of the weekly limit until November 16, plus 1 reset |
The number of resets depends on how long you have paid: 1 under 6 months, 2 from 6 to 12 months, 3 above 12. A reset gives back the Fable share of your weekly limit and refills a 5-hour session. Only accounts with paid billing before 10:00 am PT on November 9 qualify.
In the apps, pick Fable 5.2 from the model menu. In Claude Code it needs version 2.1.321 or later.
How the weekly limit and the 50% Fable cap work is in my guide to Claude Max and Ultra usage limits.
What Changes for Developers Moving to Fable 5.2?
Most code works as it is, with a few breaking details in Anthropic's release notes. Check these before you swap the model ID in production.
- Prefix check: if your code edits the system prompt, the tools or an earlier message and then sends back a Fable 5.2 thinking block, the request returns a 400 error, on every account.
- Computer use needs the computer toolset 20260801. The older computer 20251124 tool returns a 400 error everywhere except Amazon Bedrock.
- With the advisor tool in beta, a Fable 5.2 executor accepts only Fable 5.2 or Mythos 5.2 as its advisor.
- Thinking blocks work only in the account that produced them, or a linked account.
- Fast mode: set speed to fast on the request, with no beta header.
Should You Switch to Fable 5.2?
Keep Opus 5.5 as your default and move only the jobs that fail on it. Fable 5.2 costs twice as much per fresh token. My outbound agents mostly do research, drafting and data cleanup. Opus 5.5 is already good enough there. I would move only the long, messy jobs where Opus gives up or loops.
A catch for security teams. Fable 5.2 blocks most cybersecurity tasks and answers them with Opus 4.8 instead. Anthropic says it blocks about 55% less often than Fable 5.1. If security is your work, you need Mythos 5.2 through the Cyber Verification Program.
- Pick 5 real jobs from last month where Opus 5.5 struggled.
- Run each on Opus 5.5 at high effort, then on Fable 5.2 at medium and high.
- Log cost per finished job, time and how many edits you made by hand.
- Move a job to Fable 5.2 only if it wins on cost per finished job or saves real hours.
- On Pro, run the test before November 16, while Fable sits inside your limit.
Before you pick a plan for the test, check every Claude plan and price.
For every model price and the hidden fees on a Claude bill, read my guide to Claude API pricing.
And if you want agents like these working inside your sales team, that is what we build in our go-to-market consulting at Growth Cab.
Frequently asked questions
How much does Claude Fable 5.2 cost?
Claude Fable 5.2 costs $8 per million input tokens and $40 per million output tokens on the Claude API, 20% less than Fable 5.1. Cache reads cost $0.20, the Batch API $4 and $20, and fast mode $16 and $80. Mythos 5.2 has the same prices. Prices read on Anthropic's pricing page, November 9, 2026.
When was Claude Fable 5.2 released?
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026. Fable 5.2 is available on the Claude apps, Claude Code, the Claude Platform, Amazon Web Services, Google Cloud and Microsoft Azure. Mythos 5.2 goes only to organizations in Anthropic's verification programs.
Is Claude Fable 5.2 included in Claude Pro?
Normally it is outside Pro limits and runs on usage credits. For launch week, Pro and Team standard seats can spend up to half the weekly limit on Fable models at no extra cost until November 16, 2026 at 11:59 pm PT, plus 1 Fable reset. After that, Fable goes back to usage credits.
What is the difference between Fable 5.2 and Mythos 5.2?
They are the same underlying model at the same price. Fable 5.2 has safeguards that block most cybersecurity and research biology tasks, so it is open to everyone. Mythos 5.2 drops those blocks and is available only to vetted organizations through the Cyber and Life Sciences Verification Programs.
Sources
Anthropic, Introducing Claude Fable 5.2 and Claude Mythos 5.2, November 9, 2026, read November 9, 2026.
Claude Platform Docs, Pricing, read November 9, 2026.
Claude Platform Docs, Models overview, read November 9, 2026.
Claude Platform Docs, Release notes, November 9, 2026, read November 9, 2026.
Anthropic Help Center, Claude Fable 5.2 launch offers, read November 9, 2026.
Anthropic Help Center, Claude Fable models on your plan, read November 9, 2026.
