ceven

AI/ML API

Routes your data through dozens of frontier models to generate content, moderate user inputs, and automate complex reasoning tasks in real time.

Try AI/ML API in Ceven

Ask Ceven anything
Standard

Why use Ceven?

  1. AI native AI/ML API integration

    • Describe the outcome and Ceven picks the right AI/ML API calls, fills the parameters, and checks the result.
    • Structured, agent friendly tool schemas so each call runs reliably instead of by guesswork.
    • Rich coverage for reading, writing, and querying your AI/ML API data, across all 30 of its actions.

  2. Managed auth

    • Built in OAuth with automatic token refresh and rotation.
    • One place to manage, scope, and revoke AI/ML API access.
    • Per user and per environment credentials instead of shared keys.

  3. Agent optimized design

    • Actions are tuned from real success and error rates so reliability climbs over time.
    • Full execution logs so you always know what ran in AI/ML API, when, and on whose behalf.
    • The agent pauses and asks when AI/ML API is unclear instead of plowing ahead.

  4. Enterprise grade security

    • Fine grained access so you control which agents and people can reach AI/ML API.
    • Least privilege by default, read scopes first and only the writes a workflow needs.
    • A full audit trail of every AI/ML API action to support review and sign off.

Supported tools

Every action Ceven's agents can run on AI/ML API, and when to use it.

Text Chat Completion
Use this when you need to generate a response or continue a conversation. Pass the chat history and the desired model ID to get a natural language reply.
Moderation Completion
Use this to check if text or images violate safety guidelines. Run this before saving user content to a database to filter out unsafe material.
Text Completion
Generate a completion for a given prompt. Use this for single turn tasks like summarization or headline generation.
Image Generation
Create an image from a text prompt. Use this to generate blog headers or social media assets automatically.
Text to Speech
Convert written text into audio files. Use this to create voice overs for automated video workflows.
Speech to Text
Transcribe audio files into written text. Use this to process meeting recordings or voice notes for analysis.
Embeddings Generation
Convert text into numerical vectors. Use this when building a knowledge base for semantic search or RAG workflows.
Model List
Pull a list of all available models and their capabilities. Use this to verify which model IDs are currently active.
Image Analysis
Describe the contents of an image. Use this to generate alt text or categorize product photos.
Translate Text
Convert text from one language to another using a specified model. Use this for global content distribution.
Summarize Document
Condense long text into a short summary. Use this to create executive briefs from long transcripts.
Classify Sentiment
Determine if a piece of text is positive, negative, or neutral. Use this to trigger alerts for unhappy customers.

12 actions · scroll to see them all

Frequently asked questions

Ceven allows you to specify the exact model ID within the action configuration. Since AI/ML API aggregates many different providers, you can choose between the most capable frontier models for complex reasoning or smaller, faster models for simple classification. If you leave the model field blank, the agent uses a default high performance model. You can also build logic into your workflow to switch models dynamically. For example, you can use a cheap model for an initial draft and a more expensive model for the final polish. This ensures you only spend your credits on the most expensive compute when the task actually requires that level of intelligence.
AI/ML API employs different rate limits based on your account tier and the specific model you are calling. When Ceven encounters a rate limit error, the agent automatically triggers an exponential backoff strategy. It will wait for a few seconds before retrying the request, increasing the wait time with each failure. If the request fails after five attempts, the workflow will stop and send you a notification. To avoid this, we recommend monitoring your usage in the AI/ML API dashboard and upgrading your tier if you plan to run massive batch processes. You can also distribute your requests across different models to balance the load.
Yes. The moderation endpoint in AI/ML API is designed to handle both text and image inputs. When you route an image through this tool, the API analyzes the visual content for prohibited material based on safety guidelines. Ceven can be configured to act on these results immediately. For instance, if an image is flagged as unsafe, the agent can move the file to a quarantine folder and notify a human moderator. This is far more efficient than manual review for high volume uploads. You can set the sensitivity levels within the workflow to ensure you are not over filtering benign content while still maintaining safety.
Ceven stores your AI/ML API keys in an encrypted vault. We never expose these keys to the end user or the model prompts. When a workflow action is triggered, Ceven retrieves the key from the vault and injects it into the request header. You can rotate your keys at any time within the AI/ML API dashboard and simply update the key in your Ceven connection settings. This ensures that your account remains secure and that only authorized workflows can make requests on your behalf. We follow industry standard encryption protocols to ensure that your credentials never leak during the transit of a request.
While AI/ML API supports streaming for chat completions, Ceven currently processes these as complete blocks of text to ensure the downstream workflow steps have the full context before proceeding. This means the agent waits for the model to finish generating the entire response before moving to the next step in your sequence. This approach prevents the workflow from triggering subsequent actions with partial or fragmented data. If you need to show real time typing to a user, we recommend using the AI/ML API directly for the front end and using Ceven for the backend orchestration and data processing that happens after the response is finalized.
Ceven focuses on the orchestration and execution of model calls rather than the training phase. You should perform your fine tuning directly within the AI/ML API platform or through their specific training tools. Once you have created a fine tuned model and received a custom model ID, you can simply enter that ID into any Ceven action. The agent will then route all requests to your custom model instead of the base version. This allows you to leverage the power of specialized models while still benefiting from the workflow automation and integration capabilities that Ceven provides for your daily business operations.
The maximum input length depends entirely on the specific model you select via AI/ML API. Some models support a small context window of a few thousand tokens, while others support hundreds of thousands. If you send a prompt that exceeds the model limit, the API will return an error. To handle this, Ceven provides a truncation tool that can shorten your input text to fit the window of your chosen model. We recommend using the embeddings action to create a searchable index of your documents and only sending the most relevant chunks to the model. This saves on cost and prevents the model from losing focus.
AI/ML API typically follows the privacy policies of the underlying model providers they aggregate. In most professional and enterprise tiers, data sent via API is not used to train the base models. However, you should always verify the specific terms of the model you are using within the AI/ML API dashboard. Ceven does not use your prompt data or the model outputs to train our own internal models. Your data remains isolated within your account and is only passed to the AI/ML API for the purpose of generating the requested response. We ensure that your proprietary information remains private throughout the entire workflow execution.

Alternatives to AI/ML API

Other tools that solve a similar problem. Ceven supports these too, so you can switch or run more than one at once.

OpenRouter logoOpenRouterTogether AI logoTogether AIAnyscale logoAnyscale

Try Ceven on your stack

Plug Ceven on top of the tools you already run. Connect AI/ML API and the rest of your stack, describe the outcome, and its agents handle the work end to end, days of it in minutes.

Get started for free