AI News · 7 min read
Claude Fable 5.2: 20% cheaper, 30% faster, free Fable resets
Fable 5.2 drops to $8/$40 per million tokens, runs up to 30% faster than 5.1, and paid Claude plans get one to three free Fable resets based on tenure.
What Anthropic released
On November 9, 2026, Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 ↗. It's the second update to the Fable line since Fable 5 arrived in June, after Fable 5.1 in September, as 9to5Mac notes ↗. Fable and Mythos are the same underlying model. Fable adds cybersecurity and biology safeguards, so anyone on a paid plan can use it. Mythos stays limited to organizations in Anthropic's Cyber and Life Sciences Verification Programs.
The headline changes, confirmed in @ClaudeDevs' launch post ↗, are:
Cheaper: $8 per million input tokens and $40 per million output tokens, down from $10 and $50. That's a 20% cut.
Faster: outputs are generated "up to 30% faster than Claude Fable 5.1."
A new fast mode: up to 2.5 times faster token generation, at twice the price.
Free Fable resets: subscribers on eligible paid plans get one to three resets based on how long they've paid. Pro users also get a free week of Fable.
Fable 5.2 is available now in Claude apps, Claude Code (version 2.1.321 or later), the Claude Platform as claude-fable-5-2, Amazon Web Services, Google Cloud and Microsoft Foundry. On Enterprise plans it stays off until an admin enables it. Seven days after launch it's switched on automatically, unless the admin opts out first, per Anthropic's help center ↗.
The new price, line by line
Every rate in this table is from Anthropic's API pricing page ↗, per million tokens.
| Rate | Fable 5.1 | Fable 5.2 |
|---|---|---|
| Input | $10 | $8 |
| Output | $50 | $40 |
| 5-minute cache write | $12.50 | $10 |
| 1-hour cache write | $20 | $16 |
| Cache hit | $0.25 | $0.20 |
Cache hits stay at 0.025 times the input price, a quarter of the 0.1x most Claude models use. US-only inference costs 1.1 times the standard rates.
Speed: standard vs fast mode
| Mode | Speed claim | Price (in / out) |
|---|---|---|
| Standard | Up to 30% faster than Fable 5.1 | $8 / $40 |
| Fast mode | Up to 2.5x faster generation | $16 / $80 |
In the API, you turn on fast mode with "speed": "fast". In Claude Code, it's a toggle in the model picker, and ClaudeDevs says ↗ that usage is billed to extra usage credits. Anthropic set the default effort level to high on the Claude Platform and medium in the Claude apps and Claude Code.
Is it actually better?
Anthropic's own numbers say yes, and by a lot on agentic work. All scores below come from Anthropic's launch table. They are vendor numbers, not independent tests.
| Benchmark | Fable 5.2 | Fable 5.1 | Best rival |
|---|---|---|---|
| FrontierCode 2.0 | 49.1% | 32.4% | Opus 5.5: 45.5% |
| FrontierSWE v2 | 69.9% | 56.3% | GPT-6 Astra: 65.5% |
| AutomationBench | 51.4% | 31.4% | Gemini 4 Carbon: 53.9% |
| Remote Labor Index | 41.3% | 17.9% | Opus 5.5: 21.3% |
| Humanity's Last Exam | 74.2% | 65.6% | Opus 5.5: 67.7% |
| OSWorld 2.1 | 89.3% | 83.5% | Opus 5.5: 87.2% |
The cost-per-task claim may matter more than the list price. On FrontierCode 2.0, Anthropic says Fable 5.2 at High effort scored 45.3%, about level with Opus 5.5 at Max effort (45.5%), "for roughly half the cost per task." The New Stack ↗ points out it isn't a clean sweep: Gemini 4 Carbon still leads AutomationBench, and Opus 5.5 keeps a small Elo lead on the AA-Briefcase knowledge-work test (1800 vs 1785).
Anthropic also reports about 55% fewer safety blocks than Fable 5.1 in early testing, with about 60% fewer in cybersecurity and 40% fewer in biology. Blocked cyber requests are still answered by Opus 4.8, and blocked biology requests by Opus 5.
How the free Fable resets work
The details are on Anthropic's Fable 5.2 launch offers page ↗. On Max and Ultra plans, premium Team seats and premium seats on legacy seat-based Enterprise plans, the number of resets depends on how long you've been a paid subscriber:
| Paid subscription length | Fable resets |
|---|---|
| Less than six months | One |
| Six to 12 months | Two |
| More than 12 months | Three |
Tenure counts from your first paid bill, and a cancel-and-return doesn't restart the clock. On Team and Enterprise plans, it counts from the organization's first bill, so every premium seat in an org gets the same number. These resets expire on March 8, 2027 at 11:59 pm PT.
| Plan | What you get | Expires |
|---|---|---|
| Max, Ultra, premium Team and legacy Enterprise seats | One to three Fable resets (table above) | Mar 8, 2027 |
| Pro and standard Team seats | Fable included up to 50% of weekly limits through Nov 16, plus one reset | Nov 16, 2026 |
| Free, standard legacy Enterprise, self-serve and usage-based Enterprise | Nothing | — |
There are a few catches:
Billing cutoff: you need paid billing in place before November 9, 2026 at 10:00 am PT. A trial that started before then gets its resets once it converts to paid.
It's a Fable reset, not a full reset. On your weekly limit, it gives back only the usage you spent on Fable models. Anthropic's example: if you've used 80% of your week, 50% on Fable and 30% on other models, a reset takes you back to 30%. If your account has a five-hour limit, the reset refills that window completely.
It doesn't raise your limits. The cap of 50% of weekly usage on Fable models still applies.
Where to redeem: go to Settings > Usage and click "Reset for free." That button isn't in Claude Mobile or Claude Code in the terminal yet, so use the web app or Claude Desktop. If you cancel or downgrade first, you lose any unused resets.
One small mismatch: the ClaudeDevs thread ↗ names only "Ultra and Team Premium" for the tiered resets. The help center page, which is more detailed, also includes Max plans and premium seats on legacy seat-based Enterprise plans. We've gone with the help center.
What people are saying
Early reactions split between benchmark excitement and confusion about what a "Fable reset" covers:
"On FrontierCode 2.0, Fable 5.2 takes the #1 spot at a score of 49.1," posted Cognition ↗, which added the model to Devin the same day.
"On Zapier workflows, it performs 2.3x better than Fable 5.1. It's the best @Anthropic model we've tested to date," wrote Zapier CEO Wade Foster ↗.
"Fable 5.2 released and i got 3x Fable reset also, epic!" said Jaanus L. ↗
"Do the Fable resets also reset weekly usage or just the Fable usage and then Fable draws from weekly still?" asked Brian Cardarella ↗. According to the help center, it's the Fable usage only.
"we need weekly limits reset, not Fable one... we cant use it if we dont have weekly limits," replied Matt ↗ to ClaudeDevs.
We couldn't pull Reddit threads for this story, so these reactions are from X only.
Our take
For API users, Fable 5.2 replaces 5.1 outright. It costs 20% less on every token line, it's faster, and it scores higher on every benchmark Anthropic published for both. If you run Fable 5.1 in production, the only work is changing the model id to claude-fable-5-2 and retesting your prompts.
It's still not the default pick. Anthropic itself tells most users to start with Opus 5.5, which is $4 / $20 per million tokens ↗, half of Fable's new rate. The half-cost-per-task FrontierCode result is a reason to test Fable on your hardest agentic jobs. It doesn't prove Fable is cheaper for everyday prompts. Against OpenAI, GPT-6 Astra still lists at $10 / $50 ↗ for prompts up to 272K tokens, so Fable now undercuts it by 20% on list price while beating it on Anthropic's coding tests. For cheaper options, see our Claude Code alternatives guide, and for tool picks, our Claude Code review.
For subscribers, the resets are worth more than they sound, but only if you use Fable. On a Max or Ultra plan, a reset restores up to half your week, since Fable is capped at 50%. Save it until your Fable share runs out. Pro users get the better deal this week: seven days of Fable inside the plan, which normally runs only on paid credits. If you're watching Claude usage limits, our reports on the Max plan lawsuit and the Opus 5.5 "nerf" debate give more background.