Supporting teams advancing AI and science.

We’re Iacon Autonomics, a research lab in Palo Alto building ArkenOS, a workspace for AI research and experimentation.

You might have an idea you’ve been meaning to test, a baseline that needs more work, or an experiment whose results don’t quite make sense yet. We built ArkenOS to help with that work, and we’d like to give more teams the chance to use it.

Through the Iacon Grants Program, every approved team gets a full year of the ArkenOS Vector plan at no cost, a subscription worth $30,000. That covers the platform itself. On top of it, every team gets three months of uncapped AI model tokens. We’d like to help you get started and work alongside you as the project develops.

Apply

What would you like to explore?

Academic labs, nonprofits, independent researchers, and open-source teams are welcome to apply. Whether you’re testing your first idea or continuing a larger research project, we’d like to hear about it. You might be working on:

Agents, memory, tool use, and long-horizon task execution.
Fine-tuning, reinforcement learning, distillation, and model evaluation.
Robot policies, simulation, and sequential-decision systems.
AI for biology and medicine: protein design, genomics, and computational drug discovery.
AI for chemistry and materials: molecular design, reaction prediction, and discovering new materials.
AI for physics, climate, and engineering: scientific simulations, surrogate models, and experiment planning.
Benchmarks, adversarial testing, and research environments.

Your research doesn’t have to be about AI itself, or fit one of these categories. If you’re bringing AI into a field where its use is still taking shape, we’d love to hear what you want to try.

If you build test suites or scoring methods, explore our evaluation and benchmark partnerships.

A year to build, experiment, and learn

Every award pays for twelve months of the ArkenOS Vector plan, our standard $2,500 monthly team rate, with ten seats for your team. AI model tokens are covered separately below.

For the first three months from your start date, model tokens are on us, with no cap. Run the full sweep, let the long-horizon agent keep going, rerun the ablation you’d normally skip. We want you researching at full scale, so tokenmaxxing is welcome. Uncapped tokens cover Google models and open-source models from Z.ai, DeepSeek, and Kimi, served through American model providers. Models from xAI, OpenAI, Anthropic, and Meta are coming soon. If your team is awarded model credits beyond those three months, we’ll confirm the amount and eligible models in writing.

Uncapped tokens are for your team’s research in ArkenOS, under fair use. They can’t be resold, shared with people outside your team, or used to serve a product or other workloads unrelated to the project. If usage looks like it falls outside that, we’ll talk with you before changing anything.

Bring a project into ArkenOS and work with its agents to implement an approach, run experiments, and decide what to try next. When an experiment takes days or weeks, you set the direction and review progress while the agents do the work.

We’ll help you get your first project running, and you can come back to us with questions or problems along the way. We want to learn what’s working for your team and where you need more help.

Your research, your choice of models

You may already have models you like working with, or a provider your institution has approved. Your team may also have specific requirements for data retention and access. We want you to keep those choices, so you can bring your own supported model accounts into ArkenOS.

We support integrations with OpenAI, Anthropic, Google, and model services through AWS and Azure, among others. Tell us what your team uses and what you need from the setup. We’ll help you connect it and work through the access and data-handling requirements.

If your team already uses Codex, Grok, Claude, or Gemini, tell us which account or subscription you have. We’ll help you check the supported way to connect, including whether a separate API key is needed.

Tell us what you’d like to run. Whether you already have a provider or are choosing one for the first time, we’ll help you get connected so you can start experimenting.

How the grant supports your team

We want you to start knowing what support you can count on. Once your application is approved, we’ll confirm your start date, your Vector plan, your three months of uncapped AI model tokens, and any model credits after that. The platform award pays for your subscription only. Model tokens and credits pay for eligible model usage through ArkenOS.

Each award covers one approved team: the named group of researchers or builders working together in a single ArkenOS workspace, with a designated team lead, with up to ten seats. A team can receive one grant under this program, and the award can’t be transferred or shared with a separate team. Neither platform nor model grant credits can be exchanged for cash.

The award runs for twelve consecutive months from your start date, the date you chose or the day you’re approved, whichever is later, and can’t be paused. Any unused model credits expire at the end of that period. We’ll notify you before your grant-funded access ends so you can decide what comes next. We won’t automatically move your team onto a paid subscription or start charging a platform fee; continuing on a paid plan requires your choice to subscribe.

Alongside using your own supported provider accounts, you can purchase tokens through ArkenOS at subsidized rates for frontier open-source models served by American model providers. All providers serving model usage purchased through our token store operate under zero-data-retention arrangements for that usage. We’re happy to help you compare the models available for your project.

After your first three months, model usage beyond any awarded credits, optional token purchases, and compute you bring yourself are paid separately. The platform award can’t be used for those costs. The expiry of grant credits doesn’t change the terms of credits you purchase yourself.

Before you bring research data into the workspace, we’ll help you understand how the setup handles it. Provider-side zero data retention applies to the model provider’s processing; Iacon’s own data handling and retention are described in our Privacy Policy. If you bring your own provider account, its processing follows your agreement with that provider. Platform use follows our Terms and Conditions and your written grant agreement, which specifies the grant benefits and takes precedence for grant-specific subscription charges and expiry.

What you build belongs to your team. If you’d like to share what you learn, we’d be glad to write a technical case study with you. That’s entirely optional, isn’t a condition of the grant, and would only be published with your approval.

If another team comes to mind who could use this support, we’d love an introduction. They can apply for their own grant, including colleagues at the same university or organization. We’d be happy to hear what they’re working on, too.

Tell us where you’d like to start

Tell us about the project, the people working on it, and when you’d like to start. It takes a few minutes, and we read every application. Your grant year begins on the start date you choose, or on the day you’re approved if that comes later.

When your application is approved, we set up your team’s workspace and email each person on your team their own invitation.

We’d love to see what your team can do with a year to explore.