Init Intelligence Atlas

Anthropic

2 min read

Anthropic

Snapshot

  • HQ: San Francisco
  • Status: Private (foundation model provider)
  • Strategic relevance to Init Intelligence: primary frontier-tier model (Sonnet 4.6 default / Opus 4.8 complex) per ai-compute-token-cost-trends-2026; competitive surface via the $1.5B Blackstone JV.

Models (May 2026)

  • Claude Opus 4.8 — frontier reasoning tier (released May 28, 2026, model id claude-opus-4-8; supersedes Opus 4.7).
  • Claude Sonnet 4.6 — default tier for most AI ITSM tasks.
  • Claude Haiku 4.5 — lightweight tier.

Prompt caching

Strategic intel

  • Anthropic-Blackstone $1.5B JV announced May 4, 2026 — see anthropic-blackstone-jv. Anthropic is no longer a neutral model vendor for AI services — they have skin in a specific deployment game.
  • MCP 2025-11-25 spec (Linux Foundation AAIF) — agent-tool transport standard; Anthropic anchored.
  • OpenAI parallel $10B JV (openai-deployment-company) competes directly with Anthropic JV.

Strategic relevance (facts)

  • Default frontier-tier model for Init Intelligence.
  • Anthropic + Blackstone JV is the #1 competitive threat in the wiki risk register.
  • Anthropic is structurally entangled in AI services (the Blackstone JV) — no longer a neutral model vendor.
  • See responsible-ai-positioning-2026 for the “Most Transparent” customer-facing audit-trail framing.

Graph View

Backlinks