Models

One price.
Choose the model you need.

Map whatever model you want onto three tiers — Flow, Logic, and Architect. No provider contracts, no plan comparisons — the price you see below is the whole story, and you're billed only for the tokens you actually use.

DeepSeek V4 Flash
Fast, cheap, general-purpose model for chat & drafting

A lighter active-parameter model that responds instantly. Fastest and most economical for everyday chat, quick document edits, and repetitive tool calls (case-law/statute search, etc.).

Input$0.14 / 1M tokens
Output$0.28 / 1M tokens
Cache read$0.0028 / 1M tokens
  • ·Parameters: 284B total · 13B active (MoE)
  • ·Context: 1M tokens
  • ·Tool calling: Supported (parallel)
Good fit for
  • ·Everyday chat
  • ·Quick document edits
  • ·Repetitive tool calls (case-law/statute search)
  • ·Fast Q&A
MiMo v2.5
Lightweight thinking model operated by Xiaomi
ZDR (Zero Data Retention)

An alternative routing path at a price point similar to DeepSeek V4 Flash. A light, fast thinking model with 1M token context.

Input$0.14 / 1M tokens
Output$0.28 / 1M tokens
Cache read$0.0028 / 1M tokens
  • ·Parameters: 310B total · 15B active (MoE)
  • ·Context: 1M tokens
  • ·Tool calling: Supported
Good fit for
  • ·Everyday chat
  • ·Quick document edits
  • ·Repetitive automation
  • ·Fast Q&A
MiMo v2.5 Pro
A higher-tier thinking model for complex work
ZDR (Zero Data Retention)

A higher-tier model with stronger reasoning. A DeepSeek V4 Pro-class alternative routing path for multi-step brief drafting and evidence analysis.

Input$0.435 / 1M tokens
Output$0.87 / 1M tokens
Cache read$0.0036 / 1M tokens
  • ·Parameters: 1T total · 42B active (MoE)
  • ·Context: 1M tokens
  • ·Tool calling: Supported (multi-step agent chains)
Good fit for
  • ·Multi-step brief/document drafting
  • ·Bulk evidence analysis
  • ·Complex case research
  • ·Accuracy-first work
Already subscribe to Claude or ChatGPT (Codex)? Connect it and the AI runs on your own account — free and unlimited. The models above are for when you'd rather not set up a separate account.
See pricing
FAQ

Questions about models.

How do I decide which model to use?

In Settings, you map whatever model you want onto three tiers — Flow (speed), Logic (balance), and Architect (accuracy). If you don't choose, the default (DeepSeek V4 Flash) applies automatically.

Are there any hidden costs?

No. The input/output prices shown here are the whole story. Your wallet is debited at actual cost with no margin added — no subscription fee, no minimum commitment.

Can I connect my existing Claude or ChatGPT subscription instead?

Yes. Connect it and that conversation runs on your own subscription — no credit is deducted. The built-in models above are for when you'd rather not set up a separate account.

The app is free.

Download for free and pick your model right in Settings.