Pricing

Two API keys. One transparent bill.

Typhon splits its costs cleanly. The actual coding work runs on your own Claude credentials. The orchestration around it — the reasoning, planning, and conversation — runs on ours, charged at Anthropic API rates plus a flat 5%.

Our markup

Anthropic API rates, plus 5%.

The orchestration work Typhon does on your behalf runs against our backend, which is the Anthropic API behind the scenes. We pass through Anthropic's published per-token rates and add a flat 5% markup. That's the whole pricing model.

No per-seat fees. No monthly minimums. No tiered enterprise gates. Whatever Anthropic charges this month, you pay that plus 5%.

Current per-token rates by model are published on the Anthropic API pricing page.

The right model for each task.

Anthropic publishes three Claude tiers — Haiku, Sonnet, and Opus — each at a different price point. Sonnet is the workhorse; Haiku is cheaper and faster for simpler work; Opus is reserved for the hardest reasoning.

On Typhon's orchestration side, we route each unit of work to the smallest model that can handle it well. Quick classifications, short summarizations, and routine routing decisions run on Haiku. Plan drafting, ticket decomposition, and most conversation use Sonnet. Opus shows up only where its extra reasoning earns its premium.

The result: a typical orchestration token costs you a fraction of what it would if everything ran on the most expensive tier.

Bring your own Claude.

The executor at the heart of every Typhon ticket is a headless Claude Code session. Typhon does not bill you for the tokens that session burns — you connect your own credentials and the work runs on them.

Solo developers: Authenticate with your Claude Max subscription via OAuth. The executor inherits the credential and burns the same Max quota you already pay for.
Teams: Configure an Anthropic API key. Coding work bills directly to the team's Anthropic account, with usage visible on the Typhon dashboard.

Either way, the credentials never leave your machine, and you only pay Anthropic for the work Anthropic actually did.

What the orchestration side covers.

Typhon is more than a wrapper around Claude Code. The orchestrator reads tickets, builds context, decomposes work, drafts plans, asks clarifying questions, and talks with you on the dashboard. All of that runs on Typhon's own backend — and is what the 5% covers.

Usage is visible on the dashboard in real time, broken down by project and by ticket, so you can see exactly where the orchestration spend is going.

What this means in practice.

Most of the heavy token spend on a typical ticket — exploring the codebase, writing the diff, iterating on it — happens inside Claude Code, on your credentials.
Typhon's orchestration tokens are the planning, routing, and coordination layer on top. They're a smaller fraction of the total, and they are what you pay us for.
Because both halves run on Anthropic models, the cost of a ticket scales with how much real work it actually required — not with seat count or feature tier.

A note on this page.

Typhon is early. The pricing model above is what we're committing to today. If it changes as the product matures, we'll say so here — and existing customers will hear about it before anything on their bill moves.