minimax/capabilities

10 notes, read out of this brain and free to use. Each one was extracted from a source and is re-checked against its exam.

MiniMax reasoning capability

MiniMax-M models can produce intermediate reasoning ('thinking') before their final response. You control this behavior through provider options. The reasoning output is streamed through the standard AI SDK reasoning parts.

MiniMax reasoning example

Example of using reasoning with MiniMax: const { text, reasoningText } = await generateText({ model: minimax('minimax-m3'), providerOptions: { minimax: { thinking: { type: 'adaptive' } } satisfies MiniMaxLanguageModelOptions }, prompt: 'How many "r"s are in the word "strawberry"?' }); console.log(reasoningText); console.log(text);

MiniMax video model support

You can generate videos with the MiniMax-H3 model using the experimental_generateVideo function.

MiniMax video generation behavior

MiniMax-H3 generates one video per call. Generation is asynchronous — the model creates a task and polls until it completes, then returns the resulting MP4 URL.

MiniMax video resolution support

The only supported output resolution for MiniMax-H3 is 2K.

MiniMax video result URL persistence

MiniMax video result URLs are time-limited. Download and persist the video to your own storage promptly after generation.

MiniMax video generation modes

MiniMax video generation supports four modes inferred from inputs: text-to-video (prompt only), first-frame image-to-video (pass image or frameImages entry with frameType 'first_frame'), first-to-last keyframes (pass frameImages with both first_frame and last_frame), reference-to-video (pass inputReferences with images and/or videos). Frame images and references are mutually exclusive.

MiniMax video result providerMetadata

MiniMax video results include providerMetadata.minimax with the following fields: taskId (string, ID of the MiniMax generation task), videoUrl (string, the MiniMax-hosted MP4 URL, time-limited), resolvedInputs (object with imageCount number and referenceVideoUrls string array), duration (number, duration of generated video in seconds), ratio (string, aspect ratio of generated video), resolution (string, resolution tier of generated video), usage (object with totalSeconds, inputSeconds, outputSeconds for billed seconds).

MiniMax model capabilities table

MiniMax language model capabilities: minimax-m3 supports object generation (yes), tool usage (yes), tool streaming (yes), image input (no); minimax-m2.7 supports object generation (yes), tool usage (yes), tool streaming (yes), image input (no); minimax-m2.7-highspeed supports object generation (yes), tool usage (yes), tool streaming (yes), image input (no); minimax-m2.5 supports object generation (yes), tool usage (yes), tool streaming (yes), image input (no); minimax-m2.5-highspeed supports object generation (yes), tool usage (yes), tool streaming (yes), image input (no); minimax-m2.1 supports object generation (yes), tool usage (yes), tool streaming (yes), image input (no); minimax-m2.1-highspeed supports object generation (yes), tool usage (yes), tool streaming (yes), image input (no); minimax-m2 supports object generation (yes), tool usage (yes), tool streaming (yes), image input (no).

MiniMax model capabilities table

MiniMax models support the following capabilities: MiniMax-M2 supports Text Generation, Object Generation, Tool Usage, and Tool Streaming, but not Image Input. MiniMax-M2-Stable supports Text Generation, Object Generation, Tool Usage, and Tool Streaming, but not Image Input.

Give your agent this brain