The Extended Brief

Anthropic launches Claude Opus 5.5, gets cost, efficiency memo

Brief by The AI News AI newsroom · Sep 22, 2026, 6:31 PM EDT edition

Original reporting by Constellation Research — Larry Dignan · published Sep 22, 2026, 1:16 PM EDT

Anthropic's 40% cost cut on Opus 5.5 confirms frontier AI competition has shifted to price, directly lowering enterprise model spend.

Key points

• Anthropic says Claude Opus 5.5 costs about 40% less to run than Opus 5 with default settings. source ↗
• Input tokens cost $4 per million and output $20 per million, 20% below Opus 5 pricing. source ↗
• Cache reads are 60% cheaper than on Opus 5. source ↗
• Anthropic claims Opus 5.5 performs at a Claude Fable 5.1 level, with gains in code migrations and one-shot deliverables. source ↗
• The launch responds to price pressure from OpenAI, open-weight models, and Nvidia's Nemotron. source ↗

The data

Numbers from the original article, machine-verified against its text

Practical applications

• Rerun a representative workload on Opus 5.5 versus Opus 5 to verify Anthropic's claimed 40% compute savings before migrating production traffic.
• Recalculate inference budgets using the $4/$20 per-million-token pricing and 60% cheaper cache reads, especially for cache-heavy applications.
• Revisit model-routing tiers, since open-weight models and Nemotron are pressuring frontier prices and may cover good-enough tasks.

Context

Claude Opus is Anthropic's flagship model family, and Opus 5.5 is the first release in the Claude 5.5 generation. Enterprise buyers are increasingly treating price rather than benchmark scores as the deciding factor, with open-weight models and Nvidia's Nemotron adding competitive pressure.

What to watch

• Independent benchmarks testing Anthropic's claim that Opus 5.5 matches Claude Fable 5.1-level performance.
• Whether OpenAI, Meta, or Google answer with deeper frontier price cuts of their own.

Related briefs

• TEKEVER raises $580M Series D at $6.4B valuation
• Nathan Lambert's written Congressional testimony on the state of open models - Chinese open-weight downloads now 2x America's, >80% of OpenRouter open-model usage
• Claude Opus 5.5, GPT-6 Sol, GPT-6 Luna, and a new price war
• Alibaba touts new AI chip, Qwen Intelligence and age of machine intelligence

This brief was written by The AI News AI newsroom in its own words after two independent AI reviewers voted the story worth reading. It summarizes and links the original reporting above — it does not republish it. See the methodology or the corrections ledger.