// AI Platform Drift
US-only inference costs 1.1x for models after 1 February 2026
Claude Platform release notes · published 2026-11-18 20:32 UTC
WHAT CHANGEDAnthropic introduced an inference_geo parameter for data residency controls. US-only inference costs 1.1x the standard price for models released after 1 February 2026. This applies to Claude Opus 4.6 and future models.
WHY IT MATTERSEvery request using US-only inference on Opus 4.6 or later models costs 10% more than the standard multi-region price. If you route all traffic through US-only inference, your bill increases by 10% for those models.
WHAT TO DOReview whether your compliance requirements mandate US-only inference. If not, omit the inference_geo parameter to use standard pricing. If US-only is required, reforecast costs at 1.1x the published rates for Opus 4.6 and models released after 1 February 2026.
Model-reported confidence 95% (not measured accuracy)
Primary source ↗
written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
https://platform.claude.com/docs/en/release-notes/overview#february-5-2026 Thu, 05 Feb 2026 00:00:00 GMT -We’ve launched Claude Opus 4.6, our most intelligent model for complex agentic tasks and long-horizon work. Opus 4.6 recommends adaptive thinking (thinking: {type: "adaptive"}); manual thinking (type: "enabled" with budget_tokens) is deprecated. Opus 4.6 does not support prefilling assistant messages. Learn more in What’s new in Claude 4.6. +We’ve launched Claude Opus 4.6, our most intelligent model for complex agentic tasks and long-horizon work. For Opus 4.6, we recommend adaptive thinking (thinking: {type: "adaptive"}); manual thinking (type: "enabled" with budget_tokens) is deprecated. Opus 4.6 does not support prefilling assistant messages. Learn more in Claude Opus 4.6. The effort parameter no longer requires a beta header and now supports Claude Opus 4.6. Effort replaces budget_tokens for controlling thinking depth on new models. We’ve launched the compaction API in beta, providing server-side context summarization for effectively infinite conversations. Available on Opus 4.6. We’ve introduced data residency controls, allowing you to specify where model inference runs with the inference_geo parameter. US-only inference is available at 1.1x pricing for models released after February 1, 2026.
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
| triage verdict | MATERIAL at 0.92, needed 0.45 |
|---|---|
| triage reasoning | Multiple breaking changes, deprecations, new APIs, and pricing tier introduced for production model. |
| signals | deprecation announced: manual thinking (type: 'enabled' with budget_tokens) · new feature: adaptive thinking recommended for Opus 4.6 · breaking change: Opus 4.6 does not support prefilling assistant messages · API change: effort parameter no longer requires beta header · new feature: compaction API in beta · new feature: data residency controls with inference_geo parameter · pricing change: US-only inference at 1.1x pricing for models after Feb 1, 2026 |
| suppression | — |
| triage model | anthropic/claude-haiku-4.5 ($0.00000) |
| writer model | anthropic/claude-sonnet-4.5 ($0.00000) |