Anthropic Introduces Claude Opus 5.5 at a Claimed 40% Lower Cost
The new model is positioned near Claude Fable 5.1 for most work while lowering the cost of running Opus-class workloads, but the retained source does not establish the measurement method.
Edited by Tyronne Panaino
Anthropic introduced Claude Opus 5.5 on September 22, 2026, according to the company's official newsroom listing. Anthropic positions the model at the level of Claude Fable 5.1 for most work and says it costs 40% less to run than Claude Opus 5.
The material change is the combination Anthropic is claiming: Opus-class positioning with a lower operating cost than the preceding Opus model. That could matter to teams choosing a Claude model for sustained professional workloads, but the percentage remains a vendor comparison rather than an independently verified result.
What Anthropic announced
The retained official source establishes the product name, the September 22 announcement date and the high-level comparison with Fable 5.1 and Opus 5. It does not explain which workloads support the most-work comparison or how the 40% figure was calculated.
That boundary is important because model economics depend on more than one headline percentage. A buyer may need to compare input, output and cached-token prices, the number of tokens a task consumes, response speed and the quality required for a specific workflow. The newsroom summary does not provide those details, so this article does not infer them.
Why the cost claim matters
If Anthropic's comparison holds across real workloads, Opus 5.5 could narrow the trade-off between model capability and serving cost for teams that previously considered Opus 5 too expensive. The announcement may therefore affect model-selection tests for coding, analysis and other long-running work, even before independent evaluations are available.
The practical decision should still be workload-specific. The retained evidence does not show whether the claimed reduction applies evenly across tasks, usage patterns or deployment channels. It also does not establish that Fable 5.1-level performance applies to every category of work.
Evidence quality and limits
Internal confidence is medium. The source is Anthropic's own newsroom, which is authoritative for the fact that Anthropic announced the model and for what the company claims. It is not independent evidence that the performance or cost comparison will reproduce in customer environments.
The clean retained listing does not supply token prices, a model identifier, context limits, benchmark tables, safety results, detailed availability terms or third-party testing. Those omissions keep this report deliberately narrow.
What to watch next
The next useful checkpoints are clean technical documentation that specifies access, pricing and model limits, followed by reproducible comparisons on representative workloads. Independent tests should separate price per token from total cost per completed task and should state the effort settings and tools used.
Status
Confirmed. Anthropic's official newsroom lists Claude Opus 5.5 as introduced on September 22, 2026. The performance and cost statements remain vendor claims.
Sources
Update note: Last reviewed 2026-09-22. We will revise this post when clean technical documentation or independent evaluations establish the model's detailed economics and availability.
Sources
- Anthropic — Newsroom — official
Drafted with AI assistance from source briefs; reviewed for citation completeness and label accuracy.