--- description: "Answers to common questions about Verda Inference: API keys, model IDs, errors, streaming usage, rate limits, data use and where models run." revision_date: 04.11.2026 --- # FAQ ## Getting started ### What does preview mean? Verda Inference is live and ready for real workloads. During the **preview**, we're defining **SLAs** and **SLOs** for uptime and performance, usage limits apply, and details may change before general availability. ### Which key do I use for inference requests? Use an [inference API key](https://docs.verda.com/welcome-to-verda/api-credentials/#inference-api-keys), sent as `Authorization: Bearer ` to `https://inference.verda.com`. Cloud API credentials are for the Verda API at `https://api.verda.com`, such as [managing models](https://docs.verda.com/inference/models/#model-access-management). ### What do I send as `model`? The model ID from the model's page, for example `kimi-k3-1-pro`. See [Model IDs](https://docs.verda.com/inference/models/#model-ids). ### I get `403 subscription_required` The model isn't enabled for your project. Click **Enable** on the model's page. See [Enable a model](https://docs.verda.com/inference/models/#enable-a-model). ### How much balance do I need? At least $1 in your project balance to enable a model. The minimum top-up is $20. ### I get `402`, or the Console says inference is paused Your project balance is too low. Top up, and inference continues. See [Topping up in the console](https://docs.verda.com/welcome-to-verda/pricing-and-billing/#topping-up-in-the-console). ### I get `429` You reached a rate limit, or the model is busy. Wait for the `Retry-After` time, then retry. See [Handling 429 errors](https://docs.verda.com/inference/rate-limits/#handling-429-errors). ### My streamed responses have no `usage` Add `"stream_options": {"include_usage": true}` to the request. The last chunk then carries `usage`. See [Streaming](https://docs.verda.com/inference/openai-compatibility/#streaming). ### My request times out A request without streaming times out after 60 seconds. A streamed response can run for up to 20 minutes, so stream long generations. See [Timeouts](https://docs.verda.com/inference/openai-compatibility/#timeouts). ### Am I charged for failed requests? No. Requests that fail or are interrupted, including timeouts, client disconnects and streams that end early, aren't charged. ### How are reasoning tokens billed? As output tokens. See [Reasoning](https://docs.verda.com/inference/openai-compatibility/#reasoning). ### Why is `content` empty? The model used all of `max_tokens` while reasoning, before it could answer, so `finish_reason` is `length`. Raise `max_tokens`, or set a lower `reasoning_effort`. See [Reasoning](https://docs.verda.com/inference/openai-compatibility/#reasoning). ## Data and security ### Do you use my prompts to train models? No. Verda doesn't use your inputs or outputs to train any AI model. See the AI Terms of Service in Schedule 5 of our [Terms of Service](https://verda.com/terms-and-conditions). ### Do you store my prompts and responses? No. Request context is held briefly in an in-memory inference cache and is removed within 2 hours. We keep only the operational metadata needed to run the service and bill you. ### Is any data shared with the model provider? No. Verda hosts the models on its own infrastructure. No request data is sent to the model developer or any other third party. ### Can other customers see my cached context? No. The cache is a shared in-memory performance layer. It's never exposed through the API, and no customer can read another customer's cached content. ### Do you use subprocessors for inference? No. Inference runs without subprocessors. The list of subprocessors for other Verda services is in the [Trust Center](https://trust.verda.com). ### Which certifications cover Verda Inference? Verda is SOC 2 Type II, C5 and ISO 27001, 27017, 27018 and 27701 certified, and GDPR compliant. See the [Trust Center](https://trust.verda.com) for reports and details, and the [Terms of Service](https://verda.com/terms-and-conditions) for the terms that apply. ## Models and usage ### Where do the models run? On Verda GPUs in our datacenters in Europe. Each model's page lists the country. ### Can I get higher rate limits? Yes. [Contact support](https://docs.verda.com/welcome-to-verda/contact-support/) with the project, the model and the limits you need. ### Can I use embeddings, fine-tuning or batch requests? Not at the moment. Verda Inference serves Chat Completions and Responses. See [OpenAI compatibility](https://docs.verda.com/inference/openai-compatibility/#endpoints). ### Can I use my own or a fine-tuned model? Not on Verda Inference. Deploy it on [serverless containers](https://docs.verda.com/containers/overview/), or run your own stack on [GPU instances](https://docs.verda.com/cpu-and-gpu-instances/set-up-a-gpu-instance/). ### Where do I get help or send feedback? Use the chat in the [Console](https://console.verda.com/), or see [Support](https://docs.verda.com/welcome-to-verda/contact-support/).