LexxAutomates

platform updates ·

Anthropic releases Claude Opus 5.5, cutting costs 40% with Fable-level performance

Anthropic released Claude Opus 5.5 on Tuesday, September 22, 2026 — the first model in its new Claude 5.5 family, and the company’s answer to a new question: what’s the most capable model you can get at the lowest price? Anthropic says Opus 5.5 performs at the level of its top-tier Claude Fable 5.1 on most work while costing 40% less to run than Opus 5 on typical workloads, Reuters reported.

The pricing tells the story. Opus 5.5 costs $4 per million input tokens and $20 per million output tokens — 20% below Opus 5’s $5 and $25. Cache reads, which Anthropic says make up the majority of costs in agentic and coding workloads, fall 60% from $0.50 to $0.20 per million tokens. Output arrives more than 30% faster, and a new Fast mode in Claude Code and the Claude Platform pushes up to 2.5x speed at $8/$40 per million tokens.

The model ships with Fable-tier safeguards after external evaluation by METR and Frontier Design: Anthropic says Opus 5.5 is 85% less likely than Opus 5 or Mythos 5.1 to attempt to bypass its prescribed boundaries. It’s available now on the Claude Platform, Amazon Web Services, Google Cloud, and Microsoft Azure as claude-opus-5-5, with Claude Sonnet 5.5 and Haiku 5.5 due in the coming weeks.

Why it matters

The leaderboard in AI is no longer raw benchmark scores — it’s cost per completed task. Opus 5.5 undercuts Fable-class performance by 40% while matching it on most work, and it lands the same week OpenAI cut its own prices in half. For anyone running coding agents or long agentic workflows, this week quietly repriced the frontier.

Key facts

  • Launched September 22, 2026 — first model in the Claude 5.5 family (via Reuters)
  • $4 per million input tokens / $20 per million output tokens — 20% below Opus 5’s $5/$25
  • Cache reads cut 60% ($0.50 → $0.20 per million tokens); Anthropic says cache reads dominate agentic and coding costs
  • Company-reported benchmarks (flag as first-party): 40% lower cost on typical workloads, 30%+ faster output, 66.4% on Terminal-Bench 4.0
  • External evaluators METR and Frontier Design assessed it before release; 85% less likely to attempt boundary bypass than Opus 5 or Mythos 5.1
  • Available on AWS, Google Cloud, Azure, and the Claude Platform; Fast mode up to 2.5x speed; Sonnet 5.5 and Haiku 5.5 coming in the following weeks

Sources

Spot an error? We correct quickly and note it. Contact us via the contact page.

← More AI news