Anthropic: Claude Fable 5 (batch)

Claude Fable 5 (batch) is a text model from Anthropic, listed on api14 under the model id anthropic/claude-fable-5:batch. Anthropic describes it as a Mythos-class model built for autonomous knowledge work and coding. The batch label marks this variant as the batch endpoint of the model, which is the form you call when you want the same model served through a batch workflow rather than an interactive one.

The model accepts text, image, and file inputs and returns text. It has a context window of 1,000,000 tokens and can produce up to 128,000 output tokens in a single response. Reasoning is part of the model's design, and the API exposes controls for it, so you can decide how much internal reasoning effort the model spends before it writes its answer. Output is text only, which keeps the response handling simple if your pipeline expects strings rather than mixed content.

Those capabilities point to a few practical uses. Long documents, whole codebases, or large collections of files fit inside the context window, so tasks like summarising a repository, answering questions across a set of contracts, or extracting structured records from scanned pages and PDFs are reasonable fits. Image input helps when the source material includes screenshots, diagrams, or photographed documents that need to be read alongside text. Because the model is aimed at autonomous knowledge work and coding, it suits agent-style jobs where it reads material, reasons about it, and then calls tools to act on what it found.

You call it through api14's OpenAI-compatible API using the model id anthropic/claude-fable-5:batch, so existing OpenAI client libraries work with only the base URL and model name changed. The request parameters it supports are include_reasoning, max_tokens, reasoning, reasoning_effort, response_format, stop, structured_outputs, tool_choice, tools, and verbosity. In practice, tools and tool_choice drive function calling, structured_outputs and response_format constrain the shape of the reply, max_tokens and stop bound the generation, and the reasoning-related parameters let you tune how the model thinks before answering.

Keep the output ceiling in mind: 128,000 tokens is generous but finite, so very large generation jobs may need to be split across calls. If your task is purely interactive and you do not need batch processing, the non-batch form of the same model may be the better fit. The batch variant is the natural choice when you have many independent requests and can accept the batch workflow in exchange for running them together.