Anthropic: Claude Sonnet 5 (batch)

Claude Sonnet 5 (batch) is a text model from Anthropic, offered on api14 under the model id anthropic/claude-sonnet-5:batch. Anthropic describes Sonnet 5 as its most capable Sonnet-class model, with frontier performance across coding, agents, and professional work. The batch variant is the same model served through a batch processing mode rather than an interactive one, which makes it a fit for workloads where you submit many requests and collect the results rather than waiting on each call in a live session.

The model accepts text, images, and files as input, and it produces text as output. Its context window is 1,000,000 tokens, and the maximum output is 128,000 tokens. Sonnet 5 supports adaptive thinking with selectable reasoning effort levels, including low, medium, high, and max. That combination of a very large context and a large output budget means you can hand it long documents, mixed text and image inputs, or whole file collections in a single request and still get a substantial response back.

Those capabilities point to a few practical patterns. Large-context work such as reviewing long codebases, summarizing or extracting from lengthy reports, and answering questions across many documents fits naturally, especially when the documents arrive as files or scanned images. Coding and agent-style tasks follow from the model's stated strengths, and the tool and structured output support makes it usable as the reasoning step in a pipeline that calls other systems. Because reasoning effort is selectable, you can dial it down for straightforward extraction and up for problems that need more deliberation, and the batch mode suits jobs that run on a schedule or over a queue rather than in a chat window.

To call it through api14, use the OpenAI-compatible chat completions interface with the model id anthropic/claude-sonnet-5:batch. The request supports include_reasoning, max_tokens, reasoning, reasoning_effort, response_format, stop, structured_outputs, tool_choice, tools, and verbosity. In practice that means you can cap output length with max_tokens, control how much thinking the model does with reasoning_effort, request machine-readable output through response_format or structured_outputs, and define callable functions with tools and tool_choice. The include_reasoning and verbosity parameters let you decide how much of the model's internal reasoning and how much detail you want surfaced in the response.

One thing to weigh is the batch mode itself. It is designed for asynchronous, high-volume submission, so if you need a synchronous reply inside an interactive application, a non-batch deployment of the same model would be the more natural choice. The model outputs text only, so tasks that require generating images or audio fall outside its scope. Within text work, though, the 1,000,000 token context and 128,000 token output ceiling cover most long-document and long-generation cases without needing to chunk inputs.