Novita AI (MiniMax M3)

Overview

MiniMax M3 is a multimodal reasoning model from MiniMax supporting text, image, and video input with function calling and structured output. It handles a 977K token context window with tiered pricing based on input length — served via Novita AI serverless inference. This circuit provides an OpenAI-compatible chat completion endpoint backed by the minimax/minimax-m3 model.

Inputs

1. Messages (messages)

• Description: A list of messages comprising the conversation so far. For multimodal input, content must be an array of content parts (e.g., [{type: text, text: ...}, {type: image_url, image_url: {url: ...}}]).
• Type: array
• Required: Yes

2. Attachment (attachment)

• Description: Optional image attachment (blob ID or URL). When provided, the image is injected into the first message's content array as an image_url content part.
• Type: blob
• Default Value: null
• Required: No

3. Max Tokens (max_tokens)

• Description: The maximum number of tokens to generate in the completion. MiniMax M3 supports up to 131072.
• Type: number
• Default Value: 131072
• Required: No

4. Temperature (temperature)

• Description: Sampling temperature, controls randomness. Range [0.0, 2.0]. Lower values are more deterministic.
• Type: number
• Default Value: 1
• Required: No

5. Top-p (top_p)

• Description: Nucleus sampling. Range [0.01, 1.0]. Recommended to use either this or temperature, not both.
• Type: number
• Default Value: 0.95
• Required: No

6. Tools (tools)

• Description: A list of tools the model may call. Supports function-type tools.
• Type: array
• Default Value: []
• Required: No

7. Tool Choice (tool_choice)

• Description: Controls how the model selects a tool. auto or none.
• Type: string
• Default Value: auto
• Required: No

Outputs

1. Chat Completion Output (chat_completion_output)

• Description: The generated chat completion text — the model's final answer.
• Value: Accumulated from choices[0].delta.content stream chunks.

2. Reasoning Output (reasoning_output)

• Description: The chain-of-thought reasoning content — the model's thinking process, separate from the final answer.
• Value: Accumulated from choices[0].delta.reasoning_content stream chunks.

3. Chat Completion Tool Calls (chat_completion_tool_calls)

• Description: Tool calls made by the model (if any).
• Value: Accumulated from choices[0].delta.tool_calls stream chunks.

Try It

You can try this circuit directly in the Playground — no setup required.

Notes

1. Multimodal Input: MiniMax M3 supports text, image, and video input. Use the attachment input to inject an image into the first message's content array as an image_url content part.
2. Tiered Pricing: Pricing is based on input length — rates double when input exceeds 524,288 tokens. For long-context workloads, be mindful of the cost jump at the tier boundary.
3. Reasoning Always On: Reasoning is always on for MiniMax M3 and cannot be disabled.
4. Output Limit: Max output is 131,072 tokens.
5. Function Calling: Supports function calling and structured output via the tools and tool_choice inputs.

Enjoy .....!