Claude Sonnet 5.5 is now available in CRHQ

Anthropic's new Sonnet runs close to Opus 5.5 on real work at half the price, and it is faster than Sonnet 5. Now in the model picker on your CRHQ server.

Claude Sonnet 5.5 is now selectable on your CRHQ server. It is the headline of this release for a simple reason: on real work it scores close to Opus 5.5, it costs half as much, and it answers faster than the Sonnet before it.

Claude Sonnet 5.5 selected in the CRHQ model picker, next to Sonnet (latest) and Opus 5.5.

Close to Opus 5.5, at half the price

The figures below are Anthropic's own reported results, not our measurements.

Benchmark chart: Sonnet 5.5 vs Opus 5.5 vs Sonnet 5. Terminal-Bench 4.0: 70.6% vs 66.4% (Sonnet 5 not shown). CursorBench 4.0: 55.5% vs 57.8% vs 34.1%. GDPval-AA v2.1: 1844 vs 1846 vs 1449. OSWorld 2.1: 80.1% vs 81.8% vs 57.0%.

BenchmarkWhat it measuresSonnet 5.5Opus 5.5
Terminal-Bench 4.0Agentic coding in a terminal70.6%66.4%
CursorBench 4.0Coding inside a real editor55.5%57.8%
GDPval-AA v2.1Real knowledge work (Elo)18441846
OSWorld 2.1Using a computer80.1%81.8%
Humanity's Last Exam (with tools)Hard multidisciplinary reasoning64.5%67.7%

On Terminal-Bench 4.0, Sonnet 5.5 actually beats Opus 5.5. On knowledge work it lands within a couple of points. Against Sonnet 5 the jump is large: GDPval goes from 1449 to 1844, and CursorBench from 34.1% to 55.5%.

Anthropic is also clear about where Opus still wins: Opus 5.5 remains the stronger choice for complex, open-ended work that needs sustained judgment over a long stretch. Our suggestion is to make Sonnet 5.5 your everyday default and reach for Opus 5.5 when a task is genuinely hard.

Faster and cheaper

Anthropic reports Sonnet 5.5 generating output more than 30% faster than Sonnet 5, and costing up to 30% less per task because it gets the job done in fewer tokens.

Price chart per 1M tokens: Sonnet 5.5 $2.00 input and $10.00 output, Opus 5.5 $4.00 input and $20.00 output.

Per 1M tokensSonnet 5.5Opus 5.5
Input$2.00$4.00
Output$10.00$20.00
Cache read$0.20$0.20
Cache write$2.50$5.00

It has a 1 million token context window, like the rest of the current Claude family.

How to use it

In the model picker you will see two new Sonnet options:

  • Sonnet 5.5 is pinned. Your choice stays on exactly this version.
  • Sonnet (latest) always follows the newest Sonnet, which is now 5.5.

Pick either for a new session or as an agent's default. Pricing is tracked per turn in each session's Costs tab, and Claude subscription rotation, including the Fable limits, works exactly as before.

Availability

FactValue
AvailabilityEvery CRHQ server
Live since2026-09-29
ProviderClaude
Pricing$2 in / $10 out per 1M tokens; cache reads $0.20
Client action requiredNone