# Claude Mythos Pricing: $25 / $125 per Million Tokens, $100M Partner Credits | Alher Tech

> Claude Mythos lists at $25 / $125 per million tokens, 5× Opus 4.7. Founding Glasswing partners share $100M in credits. Full pricing breakdown and real-audit cost examples.

- Canonical page: https://alhertech.com/en/claude-mythos/pricing/
- Site: Alher Tech (custom software, AI agents and SEO engineering, https://alhertech.com/)
- Contact: https://alhertech.com/en/contact/

---

Claude Mythos lists at $25 per million input tokens and $125 per million output tokens, five times the rate of Claude Opus 4.7. Founding Glasswing partners receive $100M in Anthropic credits to soften the blow. This page lays out the full pricing picture, what the credits cover, and how the per-token cost actually plays out on real cybersecurity workloads.

Updated: May 9, 2026

## List price breakdown

Input: $25 per 1M tokens. Output: $125 per 1M tokens. Blended price (input + 3× output) ÷ 4 ≈ $100 per 1M tokens, roughly 5× Opus 4.7 ($25 blended), 8× GPT-5.5 ($12 blended) and around 60× DeepSeek V4. Pricing is identical across Anthropic API, AWS Bedrock, Google Cloud Vertex AI and Microsoft Foundry; only sales-negotiated commitment discounts move it.

## $100M founding-partner credits

Each of the 12 founding partners received an Anthropic credit grant, collectively $100M. The credits do not roll over; unused balance at the end of the 12-month preview window expires. Credit consumption is tracked at list price (no in-kind discount), so the practical effect is roughly 4M output tokens per partner per day before they have to pay cash.

## What the cost looks like on a real audit

A typical Glasswing engagement is a kernel sweep of a 5M-line C codebase. Inputs land around 15M tokens (chunked source plus tooling output); the model generates roughly 2M tokens of analysis, exploit drafts and patches. That is $375 input + $250 output = $625 per audit at list price. A senior security researcher with comparable output costs $25–40K per week of effort; the per-audit ROI is straightforwardly favourable.

## Why it is priced at 5× Opus 4.7

Three drivers: (1) inference cost, because Mythos is the largest model Anthropic has ever served, with active-parameter counts well above Opus 4.7; (2) demand throttling, since a high price keeps utilisation below the level that would saturate Glasswing capacity; (3) signalling, as the price tag communicates that Mythos is not a default for general workloads. Anthropic has indicated that any future general release would land at substantially lower per-token cost.

## Frequently asked questions

### Is there volume pricing for Mythos?

Yes. Beyond the founding $100M credits, Glasswing partners with sustained spend can negotiate enterprise commitments. Typical discounts land in the 15–25% range with annual minimums.

### Does the Claude Pro / Team subscription include Mythos?

No. Pro and Team plans only cover Claude Sonnet, Haiku and Opus tiers. Mythos sits on a separate enterprise contract with no consumer-grade plan.