Providers and Quotas
Almirant needs an artificial intelligence provider to power the AI Planning and AI Agents features. This section explains how to configure your provider, select models, and manage your organization's quota system.
Supported providers
Almirant currently supports the following providers:
| Provider | Available models | Recommended use case |
|---|---|---|
| OpenAI | GPT-4o, GPT-4o-mini, o1, o1-mini | General use, planning, code generation |
| Anthropic | Claude Sonnet 4, Claude Haiku | Code analysis, complex reasoning, Q&A |
Both providers offer high-quality results. If you have no preference, OpenAI GPT-4o is a good default option for general use. For tasks requiring deep code analysis, Anthropic Claude usually delivers better results.
Configuring a provider
To connect an AI provider to your organization:
- Go to Settings in the sidebar menu.
- Navigate to the AI Providers section.
- Select the provider you want to configure (OpenAI or Anthropic).
- Enter your provider API key.
- Select the model you want to use by default.
- Save the configuration.
Getting an API key
| Provider | Where to get it | URL |
|---|---|---|
| OpenAI | OpenAI Dashboard, API Keys section | https://platform.openai.com/api-keys |
| Anthropic | Anthropic Console, API Keys section | https://console.anthropic.com/settings/keys |
The API key is stored securely in Almirant, but it is only displayed once by the provider when created. Make sure to copy it before closing the provider's window.
Changing the model
You can change the default model at any time from the provider settings. The change applies to all new AI Planning and agent executions. Jobs that are already running are not affected.
Quota system
Almirant implements a per-organization quota system to control token consumption and avoid unexpected costs.
How it works
- Each organization has an assigned quota that defines the available token limit.
- Each AI Planning or agent execution consumes tokens that are deducted from the quota.
- Consumption is recorded and visible from the organization settings.
Viewing quota usage
To check your quota status:
- Go to Settings > Organization.
- Review the AI Usage section.
- You will see a summary with:
| Metric | Description |
|---|---|
| Total quota | Token limit available for the organization |
| Tokens consumed | Total tokens used in the current period |
| Remaining quota | Tokens available before reaching the limit |
| Usage percentage | Visual indicator of relative consumption |
What happens when the quota runs out
When the quota is exhausted:
- AI Planning: You will not be able to start new planning conversations. Existing conversations will show a message indicating that no quota is available.
- AI Agents: New jobs are queued in
pendingstate and will not be processed until quota is available. - Notification: Almirant notifies organization administrators when the quota is close to running out and when it is exhausted.
If you have urgent work that requires AI and your quota is exhausted, contact your organization administrator to expand the quota or wait for the period renewal.
Quota renewal
The quota is renewed according to the organization's plan. Consult with the organization administrator to learn about the renewal cycle and applicable limits.
Best practices
- Monitor consumption regularly -- Review quota usage weekly to avoid surprises.
- Choose efficient models for simple tasks -- Use lighter models (GPT-4o-mini, Claude Haiku) for Q&A or basic planning tasks, and reserve premium models for implementation and code review.
- Optimize prompts -- Clear and concise prompts consume fewer tokens and produce better results.
- Set up alerts -- Enable quota notifications to receive warnings before running out of tokens.
MCP Tools
The following tools are available via MCP to check quota status:
| Tool | Description | Main parameters |
|---|---|---|
check_quota | Checks if the organization has available quota for an operation | organizationId, estimatedTokens |
get_quota_usage | Gets the organization's quota consumption details | organizationId, period |
Example: Checking quota before execution
Tool: check_quota
Parameters:
organizationId: "organization-uuid"
estimatedTokens: 5000
Returns { available: true, remainingTokens: 45000 } if there is sufficient quota, or { available: false, remainingTokens: 200 } if there is not.
Example: Querying quota usage
Tool: get_quota_usage
Parameters:
organizationId: "organization-uuid"
period: "current"
Returns the current period's consumption details with breakdown by operation type (planning, agents) and by project.