# Claude Opus 5.5 Jumps To Top Spot On Artificial Analysis Intelligence Index, Creates 5 Point Lead Over GPT-6 Astra > Source: > Published: 2026-09-22 16:49:54+00:00 Claude Opus 5.5 has just created the biggest gap in many months between the top model and the rest of the pack on the Artificial Analysis Intelligence Index. Anthropic’s newly released [Claude Opus 5.5](https://officechai.com/ai/claude-opus-5-5-benchmarks/) has taken the top spot on the Artificial Analysis Intelligence Index, the closely watched aggregate benchmark that [tracks where AI labs stand against each other](https://officechai.com/ai/google-slips-to-10th-place-among-ai-labs-on-the-artificial-analysis-intelligence-index/). At its maximum effort setting with fallback enabled, Opus 5.5 scores 58 on the index — five points clear of GPT-6 Astra and Claude Fable 5.1, which are tied at 53, and seven points ahead of the outgoing Claude Opus 5’s score of 51. According to Artificial Analysis, Opus 5.5 brings Anthropic to parity with GPT-6 Astra on individual evaluations like Terminal-Bench 4.0 and AutomationBench-AA, while widening Anthropic’s lead specifically in agentic knowledge work. The model tops six of the ten evaluations that make up the index: Humanity’s Last Exam, where it scores 61.4% against a previous best of 59.1% set by Fable 5.1; SciCode, where it hits 66.9% against Fable 5.1’s 63.1%; and GDPval-AA v2.1, AA-Briefcase v1.1, AA-Omniscience and AutomationBench-AA. On Terminal-Bench 4.0 specifically, it scores 59.6%, level with GPT-6 Astra’s top score and 11 points ahead of Opus 5. It trails on just three sub-tests — CritPt, AA-LCR and GDP.pdf. The bigger jump shows up in agentic knowledge work. On AA-Briefcase, Artificial Analysis’s own private evaluation for professional output quality, Opus 5.5 reaches an Elo of 1,822 — 143 points ahead of Fable 5.1, and ahead on both analytical quality and how well it presents its work. Artificial Analysis says this is the first time an Anthropic model’s presentation quality has overtaken GPT-5.6 Sol’s. On GDPval-AA v2.1, a broader test of real-world professional work, Opus 5.5 posts an Elo of 1846, ahead of Fable 5.1 by 111 points and Opus 5 by 138. Pricing has also moved. Opus 5.5 costs $4 per million input tokens and $20 per million output tokens, down 20% from Opus 5’s $5/$25. Cache reads are down 60%, from $0.50 to $0.20 per million tokens — a 95% discount versus uncached input pricing, up from 90% on earlier Opus models. Cache writes drop to $5 per million tokens from $6.25. Interestingly, Artificial Analysis found that Opus 5.5 lands at roughly the same cost per task as Opus 5 despite generating 1.6 times as many output tokens to get there — Opus 5.5 at max effort uses around 119,000 output tokens per Intelligence Index task, compared to about 73,000 for Opus 5, 78,000 for Fable 5.1, and just 27,000 for GPT-6 Astra. Four of Opus 5.5’s five effort settings — medium, high, xhigh and max — land on the Pareto frontier of the intelligence-versus-cost chart, meaning they either beat or undercut every other model scoring 50 or higher on the index, including GPT-6 Astra, Fable 5.1 and Opus 5 itself. The context window stays at 1 million tokens with text and image input, unchanged from Opus 5, and the model ships with five effort levels — low, medium, high, xhigh and max — with Anthropic’s fallback mechanism switched on for all of Artificial Analysis’s test runs. The release adds to a crowded few weeks at the top of the index, with [Xiaomi’s MiMo-V2.6-Pro taking the top open-model spot](https://officechai.com/ai/xiaomi-mimo-v-2-6-pro-benchmarks/) and [StepFun’s Step 5 Preview also landing on the cost-efficiency frontier](https://officechai.com/ai/chinas-stepfun-releases-step-5-preview-beats-gemini-3-8-flash-on-performance-and-cost/) in the same window. It also comes as Anthropic leans further into [using Claude for its own internal AI research](https://officechai.com/ai/claude-is-leading-26-of-ai-research-at-anthropic-collaborating-on-over-90/), and as the company gears up for [an IPO that’s being billed as one of the biggest in tech history](https://officechai.com/ai/spacex-openai-and-anthropic-ipos-to-be-bigger-than-all-us-tech-ipos-from-1980-to-2025-combined/).