Built by Metorial, the integration platform for agentic AI.
Provider Summary
generate text completions
generate images from prompts
generate videos from text
convert speech to text
convert text to speech
moderate content safety
generate 3D models
extract text via OCR
generate text embeddings
search web in real time
Generate chat responses, images, speech audio, transcription text, embeddings, moderation classifications, and MiniMax videos through AI/ML API. Discover model IDs and capabilities before invoking a model, identify the connected account, and retrieve asynchronous results after an initial request finishes polling.
| Tool | Purpose |
|---|---|
chat_completion | Generate a response from conversation messages. |
generate_image | Generate downloadable images; existing provider URL or base64 values are retained. |
text_to_speech | Generate downloadable speech audio and return the provider URL when available. |
speech_to_text | Submit an audio URL and poll for its transcription. |
generate_embeddings | Generate numeric vectors; base64 responses also retain their original encoded value. |
moderate_content | Classify text or images with a compatible Llama Guard model. |
generate_video | Generate a MiniMax video and poll for its downloadable result. |
list_models | Discover model IDs, aliases, endpoint types, and optional capabilities, modalities, and pricing. |
get_current_user | Identify the connected account and check its balance. |
get_video_result | Check a previously submitted video task. |
get_transcription_result | Retrieve a previously submitted transcription task. |
Use an AI/ML API key from the account dashboard. The default API base URL is https://api.aimlapi.com.
Call list_models to discover a model ID and its documentation URL. Parameters vary by model: image count, image size, negative prompts, JSON mode, voice, and other settings are not universally supported. Image moderation requires a vision-capable model such as meta-llama/Llama-Guard-3-11B-Vision-Turbo.
If an asynchronous operation is still processing, use its returned generationId with the corresponding result tool. The provider retains transcription results for one hour.
This integration is licensed under the FSL-1.1.
Built with ❤️ by Metorial