Plans for skills, traces, and teams
Browse the skills registry for free. Bring your agent’s runs into DecimalAI, measure what works, and choose the capacity your team needs.
For individuals building their first agent.
Get started freeFor small teams testing and improving agent skills.
Start with CoreFor growing teams running production workloads.
Start with ProFor organizations with larger-scale requirements.
Contact salesPrices in USD. Core and Pro are monthly subscriptions. Monthly usage resets on the first of each calendar month, UTC, and is shared across your organization.
The tools to adopt skills and understand your agent’s runs.
What changes as you move between plans. Monthly allowances are marked below.
| Feature or allowance | Free | Core | Pro | Enterprise |
|---|---|---|---|---|
| Owned skillsActive skills you create or fork into your organization. | 10 | 50 | 250 | Unlimited |
| Installed registry skillsDistinct publisher skills installed through Use. A skill used across several agents counts once. | 50 | 250 | 1,000 | Unlimited |
| Traces / monthRecorded SDK traces per calendar month. | 5,000 | 10,000 | 100,000 | Unlimited |
| Included agentsAn advisory allowance for active agents: exceeding it prompts an upgrade without blocking trace ingestion. | 1 | 3 | 10 | Unlimited |
| Team membersOrganization members and pending invitations. | 1 | 5 | 25 | Unlimited |
| Regression checks / monthChecks that compare agent behavior across versions. | 50 | 200 | 1,000 | Unlimited |
| LLM-judge evaluations / monthShared allowance for LLM-judge evaluations and real-call replay. | 100 | 1,000 | 10,000 | Unlimited |
| Hosted benchmark cases / monthHosted case executions. Fully errored cases are uncharged; imported results and publish-verification runs are exempt. | 100 | 1,000 | 5,000 | 25,000 |
| Deep safety reviews / monthOn-demand LLM reviews of private skill drafts. | 10 | 100 | 1,000 | 5,000 |
| Real-call replayReplay saved LLM calls with a model. Usage draws from the LLM-judge allowance. | Not included | Included | Included | Included |
| Audit logView your organization’s audit history. | Not included | Not included | Included | Included |
Explore the details for benchmarking, training data, and team administration.
| Feature or allowance | Free | Core | Pro | Enterprise |
|---|---|---|---|---|
| Skill Rater reports / monthAutomated skill reviews. | 100 | 1,000 | 5,000 | Unlimited |
| Hosted benchmark runs / monthHosted benchmark suites across your organization; separate from the case allowance. | 10 | 100 | 1,000 | 25,000 |
| Hosted runs per skill / monthThe per-skill cap also applies within the organization’s monthly benchmark-run allowance. | 10 | 50 | 250 | 2,500 |
| Drafted skill edits / monthLLM-generated suggestions for changes to a skill. | 3 | 30 | 300 | Unlimited |
| Feature or allowance | Free | Core | Pro | Enterprise |
|---|---|---|---|---|
| Repair batches / monthBatches that apply repairs when moving between agent versions. | 10 | 50 | Unlimited | Unlimited |
| Active evergreen tasksSaved trace tasks kept active for continuous regression checks; a current total, not a monthly allowance. | 50 | 200 | 1,000 | Unlimited |
| Feature or allowance | Free | Core | Pro | Enterprise |
|---|---|---|---|---|
| SFT training rows / monthTraining dataset rows built each month; archive and replacement refunds apply. | 500 | 5,000 | 50,000 | Unlimited |
| Fine-tuning launches / monthHosted supervised fine-tuning runs. Provider credentials are required. | Not included | 2 | 5 | Unlimited |
| Feature or allowance | Free | Core | Pro | Enterprise |
|---|---|---|---|---|
| WorkspacesWorkspaces in your organization. | 1 | 3 | 10 | Unlimited |
| API keysActive API keys in your organization. | 2 | 10 | 50 | Unlimited |
| API requests / minutePer authenticated API key. Additional endpoint safeguards apply. | 60 | 120 | 300 | 1,000 |
| API request burstMaximum immediate request burst per authenticated API key before rate limiting applies. | 10 | 20 | 50 | 200 |
How allowances are counted and what happens when your needs change.
[email protected]A trace is one end-to-end agent run. It can contain several model calls and tool calls. The trace allowance counts runs, rather than the tokens inside them.
Owned skills are copies you create or fork. Installed registry skills link to a publisher’s copy. These have separate allowances. Installing the same registry skill for several agents counts once toward your organization’s installed-skill allowance.
Monthly usage is shared by your organization across its workspaces and resets at 00:00 UTC on the first day of each calendar month. Your subscription invoice date can differ. Counts such as owned skills, installed skills, team members, workspaces, and API keys are current totals and do not reset each month.
New requests that exceed an enforced allowance are rejected. For monthly allowances, upgrade or wait for the next reset. For current-total allowances, remove unused items or upgrade. Existing data remains accessible. Rejected trace uploads are not stored for delivery after the reset, and there is no automatic overage billing.
A benchmark run executes a suite of test cases. Both the number of runs and the total number of hosted case executions have allowances, and each skill also has a run allowance. A multi-case suite can use several case executions in one run. Uploaded local results do not consume the hosted case allowance.
The included active-agent count is advisory. Exceeding it does not block trace ingestion. Demo agents and the SDK connectivity-test agent do not count toward this allowance.
Manage your subscription in Settings → Billing. Upgrading increases your allowances while carrying forward usage already recorded this month. The billing portal shows the timing and billing consequences of changes or cancellation before you confirm.
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →