# Artificial Analysis scores Claude Sonnet 5.5 at 56, two points behind Opus 5.5

Artificial Analysis says Claude Sonnet 5.5 scores 56 on its Intelligence Index at max effort, two points behind Claude Opus 5.5 (max), while using the highest output-token count it has measured and costing about $7.60 per task.

Language: en
Time zone: America/Los_Angeles
Current time (UTC, cached as of): 2026-11-17T15:04:00.000Z
Query date (America/Los_Angeles): 2026-11-17

HTML: https://didcodexreset.com/news/1c5d12ad1b724379af9bdb6e.html

Content language: en
Localization state: sameLanguage

Source: Artificial Analysis · Published 9/28/2026, 11:41:39

Artificial Analysis says Anthropic has launched Claude Sonnet 5.5 with a score of 56 on its Intelligence Index at max effort. That is 18 points above Sonnet 5 and two points behind Opus 5.5 (max), placing it second on the index.

At max effort, Sonnet 5.5 reaches 64% on Terminal-Bench 4.0, compared with 60% for both Opus 5.5 and GPT-6 Astra. Artificial Analysis also reports near parity with Opus 5.5 on AA-Briefcase (1811 versus 1822 Elo), GDPval-AA (1844 versus 1846 Elo), and AutomationBench-AA (71% versus 70%), with much higher token use. It used about 193,000 output tokens per Intelligence Index task, the highest volume Artificial Analysis has measured—roughly 60% more than Opus 5.5 (max) or Sonnet 5 (max) and about seven times GPT-6 Astra (max).

On AA-Omniscience, Sonnet 5.5 scores 54% factual accuracy against 66% for Opus 5.5, with a lower hallucination rate of 47% versus 59%. It also sits about six points lower on Humanity’s Last Exam and SciCode. Token prices match Sonnet 5 at $2 per million input tokens and $10 per million output tokens, with cache writes at $2.50 and cache reads at $0.20. Artificial Analysis estimates the cost at $7.60 per task, about 50% higher than Sonnet 5, and says the model sits off its intelligence-versus-cost frontier; at lower effort, GPT-6 Astra or Sol configurations deliver equivalent performance for less, while the high-effort setting is narrowly behind GPT-6 Sol at effectively the same cost per task.

The 1 million-token context window, with image and text input, is unchanged. Five effort settings were tested, and Sonnet 5.5 fell back to Sonnet 5 in about 0.1% of Intelligence Index tasks, mainly on Terminal-Bench 4.0. The results are from a pre-release deployment that Anthropic found had a bug degrading structured-output requests. Anthropic says that bug is fixed for the public release and expects minimal change or slightly understated performance. Artificial Analysis plans to re-run the relevant evaluations.

Tags: Claude Sonnet 5.5, Anthropic, Artificial Analysis, Benchmarks

[View original post](https://x.com/ArtificialAnlys/status/2104640155843989864)
