Research

Anthropic Claude Mythos Release Brief

Quick reference for Anthropic's Claude Mythos preview (codename Capybara) gated through Project Glasswing. What a step-change frontier model with cybersecurity emphasis means for brand visibility.

By Ramanath, CTO & Co-Founder at Presenc AI · Last updated: April 2026

At a Glance

VendorAnthropic
FamilyClaude Mythos (preview)
LaunchedAnthropic previewed Claude Mythos (internal codename Capybara) in April 2026, describing it as a step change above Claude Opus 4.6 with notably stronger reasoning, coding, and cybersecurity vulnerability detection. Access is gated through Project Glasswing, a partner program limited to roughly 50 organizations.
Context windowAnthropic has not published full spec; preview-tier access reports indicate context handling consistent with the 4.x Opus family (200K standard with extended-context variants).
PricingPreview-tier access does not have published pricing. Anthropic's pattern (Opus 4.6 to Opus 4.7) suggests Mythos will price at or above current Opus rates when it reaches general availability, with extended-thinking workloads carrying premium inference cost.
Access channelsProject Glasswing partners only at preview. Roadmap pattern from prior Anthropic releases suggests broader API access via Anthropic, AWS Bedrock, Google Cloud Vertex AI, and Microsoft Foundry once general availability lands.

Notable Benchmarks

Reported step-change gains on competition math, GPQA Diamond, SWE-bench Verified, and cybersecurity benchmarks (vulnerability detection, exploit reasoning). Anthropic has emphasized cybersecurity capability as a distinguishing axis, suggesting Mythos is the first frontier model positioned explicitly for security workloads.

Strengths

Step-change frontier reasoning, distinctive cybersecurity capability, gated access creating early-adopter signal among partner organizations, alignment with Anthropic's safety-research positioning.

Limitations

Inaccessible to most teams during preview. Not yet in general production deployments. Public benchmark numbers are pre-release and not independently replicated. Anthropic pricing pattern means Mythos will be expensive at GA.

Brand-Visibility Implications

Mythos matters for brand visibility for two reasons even before general availability. First, the partner organizations in Project Glasswing are exactly the high-signal enterprise buyers (defense, fintech, security software, frontier-AI labs) that influence broader market adoption. Second, the cybersecurity emphasis means Mythos is the model that will run unattended security audits and vendor-risk assessments. Brands flagged by Mythos as low-trust (weak signing, missing security disclosures, poor SBOM coverage) will see ripple effects in enterprise procurement workflows. See Claude visibility and Claude Opus 4.7 brief.

How Presenc AI Tracks This Model

Presenc AI monitors brand visibility on Anthropic's Claude Mythos (preview) as part of continuous multi-platform AI visibility tracking. We sample Anthropic Claude Mythos across representative prompt sets daily, compare against competitor performance on the same prompts, and flag material mention-rate changes so brand teams can respond quickly when AI representation shifts.

Frequently Asked Questions

Anthropic previewed Claude Mythos (internal codename Capybara) in April 2026, describing it as a step change above Claude Opus 4.6 with notably stronger reasoning, coding, and cybersecurity vulnerability detection. Access is gated through Project Glasswing, a partner program limited to roughly 50 organizations.
Anthropic has not published full spec; preview-tier access reports indicate context handling consistent with the 4.x Opus family (200K standard with extended-context variants).
Project Glasswing partners only at preview. Roadmap pattern from prior Anthropic releases suggests broader API access via Anthropic, AWS Bedrock, Google Cloud Vertex AI, and Microsoft Foundry once general availability lands.
Mythos matters for brand visibility for two reasons even before general availability. First, the partner organizations in Project Glasswing are exactly the high-signal enterprise buyers (defense, fintech, security software, frontier-AI labs) that influence broader market adoption. Second, the cybersecurity emphasis means Mythos is the model that will run unattended security audits and vendor-risk assessments. Brands flagged by Mythos as low-trust (weak signing, missing security disclosures, poor SBOM coverage) will see ripple effects in enterprise procurement workflows. See Claude visibility and Claude Opus 4.7 brief.

Track Your AI Visibility

See how your brand appears across ChatGPT, Claude, Perplexity, and other AI platforms. Start monitoring today.