feat(deepinfra/MiniMaxAI/MiniMax-M3): add new models [bot]

[models-bot]

Auto-generated by model-addition-agent for deepinfra/MiniMaxAI/MiniMax-M3.

---

Note

Low Risk
Metadata-only addition with no runtime or auth changes; main risk is incorrect pricing or capability flags affecting billing or routing.

Overview
Adds a new DeepInfra provider definition for MiniMaxAI/MiniMax-M3, registering it as an active serverless chat model with pricing, limits, and capabilities for routing and cost estimation.

Compared to existing MiniMax entries on DeepInfra, this spec advertises a 524288 token context window, text/image/video inputs, and features including prompt caching (with separate cache-read input cost), function calling, and json output.

^{Reviewed by Cursor Bugbot for commit 9b8d780. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

/test-models

[harshiv-26]

Gateway test results

• Total: 6
• Passed: 6
• Failed: 0
• Validation failed: 0
• Errored: 0
• Skipped: 0
• Success rate: 100.0%

Provider	Model	Scenarios
deepinfra	MiniMaxAI/MiniMax-M3	success: params:stream, params, tool-call:stream, tool-call, json-output, json-output:stream

Successes (6)

deepinfra/MiniMaxAI/MiniMax-M3 — params:stream (success)

Output

The capital of France is **Paris**.

deepinfra/MiniMaxAI/MiniMax-M3 — params (success)

Output

The capital of France is **Paris**.

deepinfra/MiniMaxAI/MiniMax-M3 — tool-call:stream (success)

Output

{"location":"London"}
VALIDATION: tool-call stream SUCCESS

deepinfra/MiniMaxAI/MiniMax-M3 — tool-call (success)

Output

Function: get_weather
Arguments: {"location":"London"}
VALIDATION: tool-call SUCCESS

deepinfra/MiniMaxAI/MiniMax-M3 — json-output (success)

Output

{
  "colors": [
    {
      "name": "Turquoise",
      "hex": "#40E0D0"
    },
    {
      "name": "Coral",
      "hex": "#FF7F50"
    },
    {
      
... (truncated, 87 chars omitted)

deepinfra/MiniMaxAI/MiniMax-M3 — json-output:stream (success)

Output

{
  "colors": [
    {
      "name": "Teal",
      "hex": "#008080"
    },
    {
      "name": "Coral",
      "hex": "#FF7F50"
    },
    {
      "name
... (truncated, 91 chars omitted)

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 9b8d780. Configure here.}

[cursor]

Context window understates capacity

Medium Severity

context_window is set to 524288 (512K), but MiniMax-M3 is documented as a 1M-token model and other DeepInfra chat configs for 1M models (for example XiaomiMiMo/MiMo-V2.5-Pro and deepseek-ai/DeepSeek-V4-Flash) use 1048576. Gateways that trust this metadata may cap or reject prompts above 512K even when DeepInfra accepts the full window.

 

^{Reviewed by Cursor Bugbot for commit 9b8d780. Configure here.}