Research
Best super intelligence model? What the 2 Nov 2026 leaderboard actually shows
A saved Artificial Analysis models board from 2 November 2026. Gemini 4 Carbon (high) leads that scrape at 59.43. None of these models is superintelligence, and this site did not run the tests.
At a glance
- Type
- Research
- Topics
- super intelligence, Artificial Analysis, LLM leaderboard
- Evidence
- Third-party research
- Written by
- Adam Green

No. None of the models below is superintelligence. The table is a third-party ordering from a saved Artificial Analysis models page, scraped on 2 November 2026. This site did not run the tests. Third-party.
What does the 2 November scrape show?
The saved file has 278 rows. Intelligence is Artificial Analysis's index. Cost per task is their figure. A dash on the source page means that metric was not measured. The eight highest intelligence indexes in the file are below.
| Model | Intelligence index | Cost per task | Input / output, $ per 1M |
|---|---|---|---|
| Gemini 4 Carbon (high) | 59.43 | $7.76 | $4 / $20 |
| Claude Opus 5.5 (max with fallback) | 58.41 | $10.04 | $4 / $20 |
| Gemini 4 Carbon (medium) | 57.71 | $6.59 | $4 / $20 |
| Claude Opus 5.5 (xhigh with fallback) | 56.99 | $6.05 | $4 / $20 |
| GPT-6 Astra (max) | 56.85 | $6.24 | $10 / $50 |
| GPT-6 Astra (xhigh) | 55.26 | $3.66 | $10 / $50 |
| GPT-6 Astra (high) | 54.85 | $3.24 | $10 / $50 |
| GPT-6.1 Sol (max) | 54.51 | $1.27 | $2 / $10 |
Is this a hands-on test by this site?
No. The Copilot vs Cursor protocol on this site is still unrun. This table does not name those products, and it does not change the editor's ratings. Treat the order as someone else's measurement, dated 2 November 2026, which will have moved since.
What should you do with the order?
Use it to see which names were at the top of that board on that day, then check price and the task you actually have. A model at 59.43 is not a system that beats the best humans in every field. The benchmark reading guide lists the questions to ask before you trust a score.
Evidence in this article Third-party research Work published by researchers or institutions other than us or the vendor. We link it; we have not reproduced it.
Frequently asked questions
Is any model on this list superintelligence?
No. Superintelligence, in research, means a system beyond the best humans in every field, and none has been shown. Executive Order 14434 (91 FR 63129) uses super intelligence as a label for AI under 15 U.S.C. 9401(3). That label is not a score.
Who measured these numbers?
Artificial Analysis, on its public models leaderboard. This page copies a scrape saved on 2 November 2026. BestSuperintelligent did not rerun the benchmarks.
Which model was first on that scrape?
Gemini 4 Carbon (high), at 59.43 on the intelligence index, with a reported cost of about $7.76 per task. Claude Opus 5.5 (max with fallback) was second at 58.41.
Does a higher index mean a better tool for you?
No. The index is a third-party blend. A model can lead it and still be a poor fit for your price, context window, or task. Read the conditions before you treat the order as a buying list.
Related reading
- Category: Research & knowledge
Sources
- Artificial Analysis models leaderboard Artificial Analysis, accessed . Third-party. Figures below are from a saved page scraped that day, not a live re-fetch and not reproduced by us.
- Executive Order 14434, Inaugurating the Era of Super Intelligence (91 FR 63129) Office of the Federal Register, accessed . Primary source for the federal use of the phrase super intelligence.