Holo Models API

Our multimodal models are designed for real-world use, processing text, images, and documents, with computer-use capabilities for any digital environment.

Learn more

Start building

Holo3-35B-A3B

Holo3-35B-A3B

Fast & Open

Near-flagship accuracy at a fraction of the cost and latency. This model is fully open-source under Apache 2.0 and available to both Free-tier (10 RPM) and Paid-tier users.

Model ID

holo3-35b-a3b

Input price

$0.25

/1M tokens

Output price

$1.80

/1M tokens

$1.80

/1M tokens

Context length

65536 tokens

Max images

5 images

Input

Text + Image (JPEG, PNG, WebP)

Text + Image (JPEG, PNG, WebP)

Output

Text

Architecture

MoE — 35B, 3B active

License

Apache 2.0 (fully open)

Learn more

Hugging Face >

Holo3-122B-A10B

Holo3-122B-A10B

Flagship

Best-in-class reasoning and navigation for complex multi-step tasks across web, desktop, and mobile environments. This model is available exclusively to the Paid-tier, featuring higher rate limits for professional workflows.

Model ID

holo3-122b-a10b

Input price

$0.40

/1M tokens

Output price

$3.00

/1M tokens

Context length

65536 tokens

Max images

5 images

Input

Text + Image (JPEG, PNG, WebP)

Output

Text

Architecture

MoE — 122B, 10B active

License

Research only (non-commercial)

Learn more

Please note, after payment is made it can take up to 15 minutes for the system to process your payment and credit your balance.

Guides

Practical guides and tutorials to help you deploy the Holo Model in minutes.

Practical guides and tutorials to help you deploy Holo Model in minutes.

Overview

Introduction

Everything you need to know before you code. Get a high-level overview of the Holo Model API, core capabilities, and how our API scales with your project.

Learn more →

Tutorials

Quick start

Launch your first project in minutes. Follow our streamlined setup guide to get your environment ready and make your first successful API request right away.

Learn more →

Tutorials

Prompting with Holo3

Optimize your results. Learn the best practices for structuring inputs and tuning parameters to get the highest quality outputs from the model.

Learn more →

Frequently asked questions

Which model should I use ?

For maximum accuracy on complex, multi-step tasks — especially in novel environments — use Holo3-122B-A10B. For latency-sensitive workloads, cost-efficient automation, or well-defined tasks, Holo3-35B-A3B delivers Pareto-optimal accuracy at significantly lower cost. Both share the same API — switching requires only a model ID change on the Canvas.

Can I self-host Holo3 models ?

Holo3-35B-A3B is fully open-source under Apache 2.0. Weights are on HuggingFace and can be loaded with the transformers library. See the self-hosting cookbook for a quick start. Holo3-122B-A10B is available exclusively through this API.

How is API pricing calculated ?

We charge per million tokens processed — both input (your prompts and images) and output (model responses) count. Image tokens are determined by screenshot resolution. No hidden fees, no minimum commitments.

How do I get started with Holo3?

First, create an account on [Portal-H] (https://portal.hcompany.ai/). To start testing you can create an API key for free. Then you can add credits from the same interface. To learn more on using the API, check our documentation.

How can I contact support ?

Reach out to us at (support@hcompany.ai) specifying the nature of your request.

Where can I see the Holo3 docs?

The H Tech Hub includes simple and straightforward guides on Holo3 to help you get fully set up: https://hub.hcompany.ai/

How do I get started with Holo3?

First, create an account on Portal-H (https://portal.hcompany.ai/). To start testing you can create an API key for free. Then you can add credits from the same interface. To learn more on using the API, check our documentation.

How can I contact support ?

Reach out to us at (support@hcompany.ai) specifying the nature of your request.

Where can I see the Holo3 docs?

The H Tech Hub includes simple and straightforward guides on Holo3 to help you get fully set up: https://hub.hcompany.ai/

How does inference billing work?

Inference is billed per model and per million input and output tokens. Rates vary by model and may change over time. Refer to the pricing (hcompany.ai/terms-of-use) for current details.

How to monitor my usage?

The credit dashboard (https://portal.hcompany.ai/credits) shows your current balance and a detailed breakdown of top-ups and usage. We will continue improving the interface to provide clearer insights.

Do credits expire?

Per our terms of use (hcompany.ai/terms-of-use), we reserve the right to expire unused credits one year after purchase.

I can’t see my credits. What should I do?

Payment can take a few minutes to process. Please allow up to 15min. If your credits still do not appear, please confirm that you have been charged and that you have received a payment receipt and/or an invoice. If you have not been charged, your card may have been rejected. If so, please try again with another one. If you can confirm you have been charged and still do not see credits, please contact us at support@hcompany.ai with your purchase so we can investigate and resolve the issue.

My credit balance is negative. How is it possible?

This may happen if your balance is low and your next request uses a lot of tokens. When you submit a request, we first check your balance to allow or reject processing. However, if your remaining balance is low, the input and output tokens required to process your request may exceed your remaining balance. A negative balance should only be a few cents. If you see the negative balance increasing or have any doubts, contact us at support@hcompany.ai.

What is the refund policy?

According to our terms of use (https://hcompany.ai/terms-of-use), credits are not refundable by default. Depending on your profile and region, refund obligations may apply. Please check our terms of use to learn more.

I can’t see my credits. What should I do?

Payment can take a few minutes to process. Please allow up to 15min. If your credits still do not appear, please confirm that you have been charged and that you have received a payment receipt and/or an invoice. If you have not been charged, your card may have been rejected. If so, please try again with another one. If you can confirm you have been charged and still do not see credits, please contact us at support@hcompany.ai with your purchase so we can investigate and resolve the issue.

Why is my credit balance negative?

This may happen if your balance is low and your next request uses a lot of tokens. When you submit a request, we first check your balance to allow or reject processing. However, if your remaining balance is low, the input and output tokens required to process your request may exceed your remaining balance. A negative balance should only be a few cents. If you see the negative balance increasing or have any doubts, contact us at support@hcompany.ai.

What is the refund policy?

According to our terms of use (https://hcompany.ai/terms-of-use), credits are not refundable by default. Depending on your profile and region, refund obligations may apply. Please check our terms of use to learn more.

Do you offer a free tier?

After registering through Portal-H (https://portal.hcompany.ai/), every user is assigned to the free tier, which provides rate-limited access to our Holo3 35B model API. This rate limit is enough to test Holo-3. If you need a higher rate limit, we recommend adding credits to move to a paid tier.

What data is logged and retained?

We only record basic logging information, such as your request time and the model used, including token counts. By default, the Model API uses zero data retention and does not save your prompts or responses. Learn more on our privacy policy (https://hcompany.ai/privacy-policy)

Do you share data with other third parties?

Our API is hosted by a secure cloud provider. H Company processes its own models, so none of your inputs (prompts) or outputs are shared with any third parties. For paid accounts, we only share your registration with a secure Payment Service Provider to process payments, billing, and your credit balance. Learn more on our privacy policy.