Claude Fable 5.2
Neural Network
Claude Fable 5.2 by Anthropic: a model for demanding technical tasks with enhanced coding and reasoning capabilities.
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 1 000 000 tokens
- Max answer length
- 128 000 tokens
- Reasoning effort
- low, medium, high, extra high, max
- Capabilities
- reasoning, function calling, Batch API
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Claude Fable 5.2?
Claude Fable 5.2 is an Anthropic model for the most demanding technical tasks. Compared to version 5.1, the developer claims improvements in programming, reasoning, and image processing, with pre-release tests showing it solving previously impossible tasks. Via BotHub, the model opens in the same window as over 250 other neural networks: you can compare answers with Gemini or GPT without rewriting integrations. Access is available without a VPN or foreign card; payment is in rubles, pay-as-you-go per token, with no mandatory subscription, and balances do not expire. For teams, there is a unified OpenAI-compatible API, contracts, invoicing, EDI, an admin panel, and limits, with data protected by AES-GCM encryption and not stored. Use cases: analyzing large codebases with refactoring and debugging, architectural design and engineering trade-off analysis, function calling for agentic scenarios, analyzing diagrams, charts, and interface screenshots, preparing technical documentation, and parsing articles. When you need a second opinion, switch to another model with one click.Frequently asked questions about Claude Fable 5.2
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
claude-fable-5.2 is an Anthropic text model for handling large volumes of information: document analysis, code work and editing, reasoning and STEM tasks, and parsing uploaded images and documents. A context window of up to 1,000,000 tokens allows you to keep large materials in a single request and refer back to them during the dialogue.
The model generates up to 128,000 tokens per response — enough for long documents, large code fragments, and detailed analyses. The input context reaches 1,000,000 tokens, so voluminous source material and detailed responses fit into a single session without splitting.
Yes, reasoning before answering is supported: the model can break down a task step-by-step, check logic, and then provide the result. This significantly helps with analytics, mathematics, and code debugging, where the process of solving is as important as the final formulation.
Yes, both modes are available: function calling and structured JSON output. This allows you to integrate the model into applications where the response must be machine-readable, and to connect external tools, directories, and data sources via API.
In addition to text, the model accepts images and documents: you can upload a scan, table, or screenshot, and it will parse the content. There is no indication of audio or video support in our data, so this should be verified separately before building a workflow around it.
There are no separate quality assessments for Russian in our data, so it is best to judge based on your own tasks. The most practical approach is to test it with real examples: document analysis, text rewriting, following instructions, and working with code and Russian comments within it.