Simple, predictable pricing.
Chat is free for Pearl Fibers staff with SSO. The API is pay-per-token, billed in tiny micro-dollar increments. No subscriptions, no minimums, no surprises.
Chat plans
- Unlimited chat with Peak AI V4.3
- Vision, document generation, web search
- Conversation history
- Sign in with your Pearl Fibers account
- OpenAI-compatible
/v1/chat/completions - All three models, streaming + non-streaming
- Per-key rate limits, monthly caps, audit log
- Usage dashboard in Settings
API token rates
Per million tokens, in USD. Rates are frozen on each request — historical bills don't change when prices change.
| Model | Best for | Context | Input / MTok | Cached input / MTok | Output / MTok |
|---|---|---|---|---|---|
Peak AI V4.3peak-v4.3 |
Deep reasoning, code, analysis | 66K tokens | $5 | $0.5 | $30 |
Peak AI V4.3 Lightpeak-v4.3-light |
Default for chat, drafts, summaries | 66K tokens | $0.75 | $0.075 | $4.5 |
Peak AI V4.3 Visionpeak-v4.3-vision |
Screenshots, charts, scanned docs | 33K tokens | $5 | $0.5 | $30 |
Repeated prompt prefixes are billed at the cached rate — about a tenth of fresh input. Agents that replay a long system prompt each turn pay it automatically.
Default rate limits
Every API key starts with these limits per minute. Admins can raise or lower them per key from the admin controls.
| Limit | Default | What it counts |
|---|---|---|
| Requests / minute | 60 | Any /v1/* call (chat completions + models). |
| Tokens / minute | 200,000 | Sum of prompt + completion tokens. |
Monthly caps
Set a hard ceiling per workspace and we'll return 402 usage_cap_exceeded as soon as the month's spend crosses it. New month, fresh budget. Error reference →
Billing
Billing is settled through your Pearl Fibers account. Usage is tallied nightly into a daily rollup and pushed to billing the following day. Detailed breakdown — by model, by key, by day — is always available in your API dashboard.