Anthropic Unveils Three AI Powerhouses: Claude Opus 4.8, Sonnet 4.8, and Mythos 1

On the same day developers browsing Google Vertex AI noticed a new model identifier claude-opus-4.8, confirming that Anthropic’s major version update is imminent.

Opus 4.8 follows the April release of Opus 4.7, which already outperformed Gemini 3.1 Pro and GPT‑5.4. The upcoming version is expected to inherit Opus 4.7’s visual upgrades, pushing UI mock‑up and complex diagram recognition accuracy beyond 98 %, while delivering cleaner one‑shot code generation and stronger logical reasoning.

Earlier, a 51‑hundred‑thousand‑line TypeScript source map (≈59.8 MB) was unintentionally published to the public npm registry with version v2.1.88 because the .npmignore file omitted the .map entry. The leak revealed references to Sonnet 4.8 and Opus 4.7, with no trace of Sonnet 4.7, confirming Anthropic’s plan to skip directly to Sonnet 4.8. Community analysis expects four major upgrades in Sonnet 4.8:

Visual ability surge : inherits Opus 4.7’s visual accuracy and extends it to a cheaper model tier.

Programming ability boost : generates cleaner code with more precise instruction following.

New "X‑high" inference tier : a higher‑intensity setting that improves logical reasoning without a proportional increase in latency.

Updated tokenizer : the new tokenizer consumes roughly 30 % more tokens for the same prompt.

In parallel, the model identifier claude-mythos-1-preview briefly appeared in the Claude UI, and the source code added the strings “Claude Code” and “Claude Security”. This indicates that Mythos 1, previously described as a “dangerous” research‑only tool, is being transitioned to a developer‑facing product line focused on code generation and security.

Anthropic is also accelerating the productization of Claude Security. A new security dashboard for enterprise customers will display discovered vulnerabilities, 7‑day and 30‑day trend charts, and deeper classification analyses. The company outlines two combined offerings:

Claude Code + Mythos : a developer‑oriented security‑aware coding assistant.

Claude Security + Mythos : an enterprise platform for automated vulnerability discovery and remediation.

Meanwhile, OpenAI is poised to release GPP‑5.6 and Google is preparing Gemini 3.5 Pro for June, intensifying the race toward artificial super‑intelligence (ASI). Anthropic’s three‑track “all‑out” strategy—advancing model capability, coding proficiency, and security—signals a decisive move to commercialize powerful yet safely governed AI systems.

Reference links:

https://x.com/kimmonismus/status/2058226072596971694

https://x.com/pankajkumar_dev/status/2057832457655959664

https://www.testingcatalog.com/anthropic-prepares-mythos-1-for-claude-code-and-claude-security/