Anthropic releases Claude Fable 5.2 and Mythos 5.2: API pricing at $8 and $40 per million tokens
Anthropic has released two new models in the Claude family: Fable 5.2 and Mythos 5.2. The company says they are better at coding, reasoning and understanding images. Fable is already available on paid plans and cloud platforms, while Mythos requires separate approval through cybersecurity or biology verification programs. API pricing for both is $8 per million input tokens and $40 per million output tokens.
AI-processed from TG: @neyr0graph (NeuroGraph); edited by Hamidun News
Anthropic has updated its Claude lineup with two models at once: Fable 5.2 and Mythos 5.2. The company promises better performance in programming, reasoning and image understanding. The announcement includes no independent measurements, so these remain the developer's claims for now. Access terms and pricing, however, are detailed, and those will determine which option offers you better value.
Who gets what
Fable 5.2 is the workhorse model. It is already available on paid Claude plans, on Claude Platform and on major cloud platforms. Getting access requires no special formalities.
Mythos 5.2 works differently. Access requires separate approval through verification programs for cybersecurity or biological research. The announcement does not explain the reasons, but the apparent rationale is clear: these are areas where a powerful model can be both highly useful and risky, so access is subject to screening. If you are building a conventional SaaS product, you probably will not need Mythos or get access to it.
Anthropic's use cases are practical. For images: analyze a reference image's lighting, composition and materials, then turn the observations into prompt revisions. For video: break an idea down into scenes and shots, describe camera movements and prepare prompts for each shot. Individual frames can be submitted for analysis of visual inconsistencies, but that does not amount to checking motion throughout the video. For graphics and 3D: write code for animated typography or a Blender scene, then check the result in the application itself. For production workflows: scripts to sort assets, assemble previews and perform repetitive file operations.
An important caveat: the model itself does not generate any images or video. Connecting an image or video generator requires a separate tool and payment for it.
Plans and API: where Fable costs more
Here is the breakdown for subscribers.
- Free — free of charge, but Fable is unavailable.
- Pro — $20 a month or $200 a year (about $16.67 a month). Fable is generally paid for with additional usage credits.
- Max — $100 a month. Fable is included within 50% of the weekly limit, with an overall usage allowance 5 times that of Pro.
- Ultra — $200 a month. Fable is likewise limited to 50% of the weekly allowance, total usage is 10 times that of Pro, and verified accounts have no five-hour limit.
The practical takeaway is straightforward. If you use Fable occasionally, Pro plus credits may be cheaper. If you use it daily for large tasks, Max or Ultra look more sensible because Fable is already included in the allowance. Keep in mind, though, that it can account for only up to half of your weekly allowance; the rest must be used on other models.
For developers, both models cost $8 per million input tokens and $40 per million output tokens. Fast Mode doubles the price to $16 and $80. These are separate expenses, unrelated to the subscription.
Consider an example. A code review agent reads 50,000 tokens of diffs and context and writes 5,000 tokens of feedback. That costs $0.40 for input and $0.20 for output, or about $0.60 per run. In Fast Mode, it is about $1.20. At 100 reviews a day, the difference becomes significant: $60 versus $120. This is our calculation based on the published rates, not figures from the announcement.
What to do
Do not switch blindly. Take 10–20 real tasks from your product, run them through Fable 5.2 and your current model, and compare quality and cost. If the improvements in coding and reasoning hold up on your data, then work out the subscription or API budget. For Fast Mode, calculate separately whether the speed justifies paying twice as much.
If you need Mythos, start the verification process early: approval takes time, and it is too soon to base customer commitments on it. When choosing a subscription, focus on Fable's access terms rather than high-profile model names: Pro requires an extra payment, while Max and Ultra include it within a limited subscription allowance.
- Nov 12, 2026AWS Opens 23 Ready-Made GenAI Components: DoorDash and Ricoh Cut Development Hours by 90%
- Nov 12, 2026Agent Builds a 4K Blender Video and Renders It on 10 GPUs in 51 Minutes for $5.75
- Nov 12, 2026Anthropic Releases Claude Fable 5.2: Twice the Price of Opus 5.5, but Cheaper Than 5.1
- Nov 12, 2026Anthropic ported Claude to AMD chips in three days using its Mythos model
Key Questions Answered
Who gets what?
Fable 5.2 is the workhorse model. It is already available on paid Claude plans, on Claude Platform and on major cloud platforms. Getting access requires no special formalities. Mythos 5.2 works differently. Access requires separate approval through verification programs for cybersecurity or biological research. The announcement does not explain the reasons, but the apparent rationale is clear: these are areas where a powerful model can be both highly useful and risky, so access is subject to screening.
What are the key details on Plans and API: where Fable costs more?
Here is the breakdown for subscribers. Free — free of charge, but Fable is unavailable. Pro — $20 a month or $200 a year (about $16.67 a month). Fable is generally paid for with additional usage credits. Max — $100 a month. Fable is included within 50% of the weekly limit, with an overall usage allowance 5 times that of Pro. Ultra — $200 a month. Fable is likewise limited to 50% of the weekly allowance, total usage is 10 times that of Pro, and verified accounts have no five-hour limit.
What to do?
Do not switch blindly. Take 10–20 real tasks from your product, run them through Fable 5.2 and your current model, and compare quality and cost. If the improvements in coding and reasoning hold up on your data, then work out the subscription or API budget. For Fast Mode, calculate separately whether the speed justifies paying twice as much. If you need Mythos, start the verification process early: approval takes time, and it is too soon to base customer commitments on it.
Test Frontier LLMs & Autonomous Agents in One Workspace
Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.