Anchor 3.0-mini
$0.002per message
Detection only. Flags every message that fails a rulepack.
- →All 13 regulatory rulepacks
- →Detection only. No rewrites.
- →No custom rules supported
One charge per message on any model, with every rulepack, rewrite, and verification included. Sign up, add a card, and start with $10 in free credits.
Three models. One meter.
Every message of up to 2,000 tokens is one charge. Pick how deep enforcement goes, and switch on any request.
Anchor 3.0Flagship
$0.01per message
Checked, rewritten if needed, and verified. Your own rules LoRA-trained into the model weights.
- →All 13 regulatory rulepacks
- →Rewrites and verification included
- →LoRA-trains your rules into the model weights. First 5 free, then $10 per rule.
Anchor 3.0-max
$0.05per message
Everything in Anchor 3.0, plus your policy docs enforced as written. No training needed.
- →All 13 regulatory rulepacks
- →Rewrites and verification included
- →Upload your policy or rules and run them as written. No LoRA training.
//01Sign up and add a card.
Your account opens with $10 in free credits. Nothing is charged until they run out.
//02Send messages through the API.
Each message is one charge at the price of the model that handled it. Rewrites and verification are part of it.
//03Add your own rules when you need them.
Every 10 rules add one message charge. That is the only way the bill grows.
Compare the models.
Same API, same rulepacks. The model decides what happens to a message and how your own rules run.
Anchor 3.0-mini$0.002 / message
Anchor 3.0Flagship$0.01 / message
Anchor 3.0-max$0.05 / message
Enforcement
Price per message (up to 2,000 tokens)
$0.002
$0.01
$0.05
What happens to each message
Checked. Failures are flagged.
Checked, rewritten if needed, verified.
Checked, rewritten if needed, verified.
Rewrites and verification
✓Included
✓Included
Free credits to start
$10
$10
$10
Policy model training
How your own rules run
Not supported. Managed rulepacks only.
LoRA-trained into the model weights. Your rules run like the built-in rulepacks.
Read from your uploaded policy or rules at run time. No LoRA training.
Setup cost
First 5 rules free, then $10 per rule trained or retrained.
$0
Cost of applying your rules
+$0.01 per message for every 10 rules.
+$0.05 per message for every 10 rules.
Deployment
Cloud, pay as you go
✓
✓
✓
13 rulepacks. Included with every model.
Every regulation pack, on every model, at no extra cost. Scope any API key to the packs it needs.
//01Financial services
- US Securities and Investment Communications
- US Banking and Lending
- US Futures and FX
- Digital Assets
//02Marketing and promotions
- US Marketing Channel
- UK Financial Promotions
- EU Promotions
- APAC and Canada Promotions
//03Insurance and real estate
- US Insurance
- US Real Estate and Fair Housing
//04Conduct, privacy and AI
- Core Conduct
- Cyber and Privacy for Financial Services
- AI Governance
More rulepacks coming soonNeed one sooner? Talk to Sales →
Included with every account.
The full platform from the first message.
//01One API for every surface
Enforce emails, chat, documents, and LLM outputs with one call, inline or as a queued job.
//02OpenAPI spec
Generate a Python or TypeScript client, or call the API from any language.
//03Policy import and MCP
Upload a policy doc, or connect Notion, Linear, Confluence, or Google Drive. ZeroDrift extracts the rules for you to approve.
//04Command dashboard
Every decision with the rules behind it, and an evidence ID for your audit trail.
//05Scoped API keys
Separate validation and admin keys, each limited to the permissions and rulepacks it needs.
//06Enterprise security
SOC 2 Type II, ISO 27001, GDPR, and HIPAA. Per-tenant isolation with customer-managed keys.
Questions developers ask before they sign up.
What counts as a message?One message of up to 2,000 tokens, checked against your rulepacks and policies, rewritten if needed, and verified. That is one charge at your model's price.
Can I switch models?Yes. Set the model on each request, switch any time, or run different models for different tasks. Each message bills at the price of the model that handled it.
Do rewrites cost extra?No. On Anchor 3.0 and Anchor 3.0-max the rewrite and its verification are part of the same charge. Anchor 3.0-mini is detection only.
Are my rules trained in, or read at run time?Anchor 3.0 LoRA-trains your rules into the model weights: the first 5 are free, then $10 per rule trained or retrained. Anchor 3.0-max reads your uploaded policy or rules at run time, with no LoRA training. On both, every 10 rules add one message charge.
What happens after the $10 in free credits?Usage is billed to the card on file at the same per-message prices. There are no seats, minimums, or platform fees.
When should I talk to sales?For in-VPC deployment or volume pricing. Everything else starts with a signup.