Mistral: Ministral 3 8B 2512 (batch)

Mistral: Ministral 3 8B 2512 (batch) is a text generation model from Mistral, part of the Ministral 3 family. Mistral describes it as a balanced model in that family: a powerful, efficient tiny language model with vision capabilities. The model id on api14 is mistralai/ministral-8b-2512:batch, and the batch suffix indicates this variant is served through a batch endpoint rather than a synchronous one, which suits workloads where you submit work and collect results rather than waiting on each call.

The model accepts text and image input and produces text output, so it can handle both plain language tasks and prompts that include images. Its context window is 262144 tokens, and the maximum output is 209715 tokens. That combination of a large context and a large output budget means you can feed in long documents or long image-plus-text prompts and still leave room for a substantial generated response. Supported request parameters include temperature, top_p, frequency_penalty, presence_penalty, max_tokens, stop, seed, response_format, structured_outputs, tool_choice and tools.

Those capabilities point to a few practical uses. Long-context text work such as summarising or extracting from lengthy documents fits the 262144 token window. Because images are accepted as input, tasks like describing an image, answering questions about a screenshot, or extracting text and structure from a scanned page are within scope. Tool calling and structured outputs make it usable for pipelines that need the model to return JSON or invoke functions, and the seed parameter helps when you want repeatable results across runs.

To call it through api14, use the OpenAI-compatible chat completions interface with the model id mistralai/ministral-8b-2512:batch. You can pass the usual chat request fields along with the parameters listed above: temperature, top_p, frequency_penalty, presence_penalty, max_tokens, stop, seed, response_format, structured_outputs, tool_choice and tools. Because this is the batch variant, plan for asynchronous submission and retrieval rather than immediate responses, and check the live price table on the page for current rates.

One limit worth noting is output modality: the model reads images but returns only text, so it is not a fit if you need generated images. If your task is purely text and does not need vision or the very large context window, a smaller or non-batch model may respond faster and with less operational overhead.