Native model freedom

Use the AI access you already pay for.

Run hosted Hermes or OpenClaw agents with Platform Credits, your provider API keys, or supported customer-owned subscriptions. Keep model choice and account failover inside the native harness where they belong.

Agent Teams hosting is still billed normally. Subscription availability and quota are controlled by each provider and native harness route.

Native account stack

Choose cost, quality, and fallback

1

Primary subscription

Use existing eligible quota first

2

Second named account

Native failover when supported

3

Provider API key

Direct provider billing when selected

4

Platform Credits

Simple prepaid access when selected

Subscription sign-in stays in the native harness. AI Agent Store does not store those OAuth tokens.

One agent, three payment routes

Choose the access route that fits each agent.

Start fast, optimize costs later, and keep different agents on different providers without rewriting their setup files.

Fastest start

Platform Credits

Launch without opening a separate provider account. Usage is deducted from your prepaid AI Agent Store balance at the published model rate.

  • One balance across supported models
  • Clear per-agent usage
  • Stop when the balance is depleted

Direct provider billing

Your API keys

Save provider keys once, assign them to the agents that need them, and let the native harness discover the models available to that account.

  • Encrypted credential storage
  • Different providers for different agents
  • Provider charges you directly

Use an eligible plan you already have

Customer-owned subscriptions

Sign in from the native Hermes or OpenClaw interface when the provider and harness support subscription access. AI Agent Store does not collect the OAuth tokens.

  • Multiple named accounts where supported
  • Native priority and quota-aware failover
  • No silent fallback to Platform Credits

Multiple accounts, native control

Stretch eligible subscription value without inventing a second router.

Connect more than one account for a supported provider, give each one a clear name, and set its priority in Hermes or OpenClaw. The harness decides when an eligible fallback is needed; AI Agent Store only hosts the runtime and shows the result.

Compatibility matters: a consumer chat plan is not automatically usable by agents. The provider and native harness must expose an authorized subscription route.

Currently supported native routes

OpenAI / Codex
MiniMax
Alibaba / Qwen
Z.AI
Kimi
Eligible xAI accounts

The list can grow as Hermes, OpenClaw, and providers add compatible native authentication routes.

Switch models at any time

Change the model, not the agent.

Select another available model in the native Hermes or OpenClaw UI. New native sessions use the new choice without restarting the whole agent, while existing sessions keep their original context and model.

Agent Teams observes the actual model requested for accurate Platform Credit billing and status. Native subscription or BYOK usage remains with that provider account.

Native software updates

Update Hermes or OpenClaw with one button.

Agent Teams checkpoints the agent, applies the latest supported runtime image, restarts it, and verifies that the native interface comes back. When a previous version is available, you keep a rollback path.

  • No manual package commands
  • Saved native state restored after replacement
  • Health verification and rollback visibility

Clear answers before you connect

Model access FAQ

Can every consumer AI subscription be connected?

No. Subscription access depends on the provider plan and the native Hermes or OpenClaw authentication route. Agent Teams shows the native options that are actually available instead of treating every chat subscription like an API plan.

Can I connect more than one subscription account?

Yes, where the native harness supports that provider route. Name the accounts, set their priority, and let the harness use eligible quota-aware failover. The harness—not AI Agent Store—owns that selection logic.

Do I need to restart an agent after changing its model?

Normally, no. The new model applies to new native sessions. An existing conversation can keep the model it started with, so begin a new session when you want the switch to take effect immediately.

What happens when a subscription reaches its limit?

If you configured another eligible account, the native harness can move to the next account according to its priority and failover behavior. Agent Teams does not silently convert the request into paid Platform Credits.

Does AI Agent Store control how the agent reasons or calls tools?

No. Hermes and OpenClaw keep control of conversation context, planning, tools, schedules, and subagents. AI Agent Store provides hosting, secure access, checkpoints, communications, and billing visibility.

Start with the access you prefer

Your agent should not be locked to one billing route.

Launch with Platform Credits, then connect eligible subscriptions or provider keys when you are ready to optimize cost and model choice.

Open Agent Teams