Skip to content
Open in ChatGPT Open in Claude

AI Units pricing

AI-powered features in Coralogix — Olly, Olly Mini (AI-powered helpers across the platform such as query assistant, Explain log, and case analysis), and AI Center evaluations and guardrails — are billed in AI Units. Every Coralogix user has access to AI features. Pay only for what you use.

AI Units are a separate pool from the units used for Logs, Metrics, and Traces. The two pools are not interchangeable.

Free monthly allowance

Two allowances apply on free plans:

  • Per user: each user receives 10 AI Units free per month for per-user AI usage.
  • Per team: teams on the free plan have an additional cap of 100 AI Units per month across all members.

Both allowances reset on the 1st of each month. When either allowance is exhausted, all AI features are blocked until the next reset.

Paid plans include additional capacity beyond the free allowance. See your plan for the details that apply to your account.

What's billed in AI Units

AI Units cover any Coralogix feature that uses an AI model to process your data. As the AI Units billing model rolls out across the platform, the following features consume AI Units:

All AI features are billed at per-model, per-token rates. Two groups of features consume AI Units, differing in who is billed:

Per-user AI features (billed per user):

  • Olly chat and scheduled tasks
  • Olly Mini: AI-powered helpers embedded across the product, including:
    • Query assistant — natural-language query suggestions in Explore
    • AI-driven case analysis in Cases
    • Explain log — natural-language explanations for individual log entries
    • AI explanations of Continuous Profiling flame graphs
    • RUM error AI analysis

Team-level AI features (billed per team):

Sending AI traces to Coralogix is billed as standard trace ingestion — there is no additional AI Center cost on top. Additional cost applies only when evaluations or guardrails are enabled.

How AI Units are calculated

AI Units are calculated per token, with rates that vary by model (or by feature type, for Eval and Guardrail) and by token type.

Every AI interaction involves different types of token processing:

  • Input: Tokens sent to the model, including your query and any retrieved context
  • Cache Write: Tokens written to the model's prompt cache for reuse in subsequent requests (supported models only)
  • Cache Hit: Tokens retrieved from the prompt cache instead of being reprocessed, reducing cost
  • Output: Tokens generated by the model in its response

Rates per model — a dash (-) indicates the model does not support that token type:
ModelInput (units/Mtok)Cache Write (units/Mtok)Cache Hit (units/Mtok)Output (units/Mtok)
GPT-5 mini0.77-0.086.15
GPT-5.4 mini2.31-0.2313.85
GPT-5.13.85-0.3830.77
GPT-5.25.39-0.5343.07
GPT-5.47.7-0.7646.15
Claude Haiku 4.53.083.850.3115.38
Claude Sonnet 4.5/4.69.2311.540.9246.15
Gemini 3.1 Pro6.16-0.6136.92
Gemini 3 Flash1.54-0.153
Eval2.31-0.2313.85
Guardrail2.31-0.2313.85

Example

A query using Claude Haiku 4.5 that processes 50,000 input tokens and generates 2,000 output tokens costs:

  • Input: 0.05 Mtok x 3.08 units/Mtok = 0.154 units
  • Output: 0.002 Mtok x 15.38 units/Mtok = 0.031 units
  • Total: 0.185 AI Units

In AI Center graphs and tables, policy names include their type — for example, Toxicity (Eval) or Prompt Injection (Guardrail) — so you can distinguish between them at a glance.

Monitor your usage

Track AI Units consumption on the AI Units tab of the Data Usage page. The tab has three stacked sections:

  • Total: combined daily AI Units across per-user and AI Center usage
  • Olly: per-user breakdown for Olly usage, with the free-tier banner on free teams
  • AI Center: policy-level breakdown for evaluations and guardrails

AI Units consumption is tracked separately from your Logs, Metrics, and Traces quota. Two admin controls apply:

  • Team-wide limit: use Quota Rules to cap the team's total AI Units consumption on the olly entity type.
  • Per-user and automation limits: use Olly usage management to cap individual users and automations.

Next steps

Learn how usage is calculated, monitored, and reported across all pillars in Data usage.

Need help? Contact Support.
What's new? Find out here.
LLM? Read llms.txt.
Previous Quota Rules
Next Data usage