Anthropic

Entity page for Anthropic — the AI lab behind Claude and a 2026 Frontier Trinity leader.

Overview

Anthropic develops frontier language models with an emphasis on alignment and safe deployment. In this knowledge base, Anthropic appears primarily through the Claude product line and Claude Code agentic coding tool.

Key products (2026)

Product	Type	Notes
Claude Opus 4.8	Frontier model	Trustworthiness, math, SWE-bench Pro
Claude Fable 5	Mythos-class model	Long-horizon coding; public with safeguards
Claude Mythos 5	Gated model	Same capability, restricted access
Claude Sonnet 5	Agentic mid-tier	$2/M input pricing
Claude Code	Coding agent	Gartner 2026 MQ Leader; ~10% of public GitHub commits
Claude Science	Research workbench	60+ scientific tools, multi-agent review, native 3D rendering, drug discovery
Claude Haiku 4.5	Efficient model	Used in CLAW-00 OpenClaw setup

Strategic positioning (2026)

Anthropic's frontier thesis: "Make the model trustworthy for autonomous work" — optimizing honesty, math reliability, and agentic coding over raw benchmark maximization across all domains.

Distinct from:

• OpenAI — terminal workflows, SWE-bench Verified
• Google — MCP tool orchestration, multimodal workflows

→ Frontier Models

Enterprise recognition

• Gartner Magic Quadrant 2026 — Leader (Enterprise AI Coding Agents) for Claude Code
• First-time Leader alongside GitHub, OpenAI, Cursor

→ Gartner Magic Quadrant Enterprise Ai Coding Agents 2026 05 26

Notable releases timeline

Date	Release
May 2026	Opus 4.8 — honesty, Dynamic Workflows
Jun 2026	Fable 5 / Mythos 5 — Mythos-class tier
Jun 12–30, 2026	Export control suspension (19 days) — Amazon jailbreak report, global suspension, negotiated resolution
Jul 1, 2026	Fable 5 restored globally with enhanced safeguards, usage-credits pricing
Jul 7, 2026	Usage-credits pricing takes effect ($10/$50 per million tokens)
Jun 30, 2026	Claude Science launched — AI workbench for drug discovery and biomedical research
Ongoing	Claude Code enterprise adoption

Export control episode (June–July 2026)

The Fable 5 / Mythos 5 launch was interrupted by the most significant government intervention in frontier AI deployment to date:

• June 12: US Commerce Department issued export control directive after Amazon researchers reported a jailbreak technique
• June 12–30: 19-day global suspension; government coordination and safeguard improvements
• June 30: Export controls lifted; Mythos 5 partially restored to ~100 US organizations
• July 1: Fable 5 restored globally with enhanced safety classifier (99%+ block rate), 30-day data retention, and Opus 4.8 fallback
• Key outcome: Industry's first shared jailbreak severity framework co-developed with Amazon, Microsoft, and Google

→ Claude Fable 5 Mythos 5 Redeployment Export Control Lifted Safeguards Industry Framework 2026 07 02

Key articles

Topic	Article
Opus model line	Claude Opus
Opus 4.8 analysis	Claude Opus 4 8 Agentic Coding Honesty Dynamic Workflows 2026 05 28
Fable 5 breakthrough	Claude Fable 5 Mythos 5 Analysis 2026 06 10
Fable 5 / Mythos 5 redeployment	Claude Fable 5 Mythos 5 Redeployment Export Control Lifted Safeguards Industry Framework 2026 07 02
Claude Science workbench	Claude Science Ai Workbench Drug Discovery Biomedical Research 2026 07 07
Enterprise showdown	Enterprise Ai Coding Agents Showdown Claude Codex Cursor Github 2026 05 27
Mythos preview	Mythos Preview Research Summary
API pricing / costs	Ai Coding Agent Monthly Cost Claude Api

Related

• Concepts: Agentic Coding, Frontier Models
• Infrastructure: Openclaw (uses Claude Haiku by default)