🔍 Read the full analysis: Gemini 4 Carbon Leads The AI Rankings By One Point—Is That Enough? on ThorstenMeyerAI.com
Get hardware and tech essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Gemini 4 Carbon scored 59 on Artificial Analysis Intelligence Index v5, one point ahead of Claude Opus 5.5 and two ahead of GPT-6 Astra. It placed first in four of nine tests, but trails competitors on other tasks, and Google has not made it publicly available. The ranking reflects a specific benchmark mix, not a universal measure of model quality.
Artificial Analysis rebuilt its leaderboard on October 27, introducing version 5 of its index. The reported scores place Claude Opus 5.5 at 58, GPT-6 Astra at 57, GPT-6.1 Sol at 55 and Google’s earlier Gemini 4 Argon at 50. Carbon’s result refers to its high-reasoning setting; the rival scores cited are for the settings named in the source report.
Carbon finished first in four of nine tests: HLE-Diamond, Terminal-Bench-Science, AA-Omniscience and AutomationBench-AA. The source says it was weaker than leading competitors in office-style knowledge work, real-world coding, physics research, terminal tasks and document reasoning. On HLE-Diamond it scored 69%, against 68% for Opus 5.5; on Terminal-Bench-Science it scored 65%, against 63% for GPT-6 Astra.
The report estimates Carbon’s cost at $7.76 per index task at standard prices. That is below the cited $10.04 for Claude Opus 5.5 but above GPT-6 Astra’s $6.24 and GPT-6.1 Sol’s $1.27. It also reports introductory API pricing of $2 per million input tokens and $10 per million output tokens, later rising to $4 and $20 respectively. Those are reported prices; actual costs depend on usage and task requirements.
Infographic 01 · What the #1 is made of
Number one overall. First in four of nine tests.
The Intelligence Index is a decathlon score: nine tests, one number. Carbon wins on reasoning, science and careful knowledge, and trails on office work and coding.
The nine tests behind the score
Carbon at its highest setting (high) against the best model on each test.
| Test | Weight | Carbon | Leader or runner-up |
|---|---|---|---|
| HLE-Diamond Hard reasoning + knowledge questions | Weight 10% | Carbon 69% | Carbon leads Next: Claude Opus 5.5, 68% |
| Terminal-Bench-Science Science research work in a terminal | Weight 15% | Carbon 65% | Carbon leads Next: GPT-6 Astra, 63% |
| AA-Omniscience v1.1 Knowledge, wrong answers penalized | Weight 15% | Carbon 54 | Carbon leads Next: Claude Opus 5.5, 48 |
| AutomationBench-AA Business workflows across apps | Weight 5% | Carbon 78% | Carbon leads Next: Gemini 4 Argon, 78% |
| AA-Briefcase Office-style knowledge work | Weight 15% | Carbon 1,609 Elo | Behind Claude Sonnet 5.5, 1,812 |
| FrontierCode 2.0 Code a maintainer would merge | Weight 15% | Carbon 41% | Behind Claude Opus 5.5, 46% |
| CritPt v2 (hard) Research-level physics | Weight 10% | Carbon 63% | Behind GPT-6 Astra, 71% |
| Terminal-Bench 4.0 Agentic terminal tasks | Weight 10% | Carbon 65% | Behind Kimi K3.1 Pro, 67% |
| GDP.pdf Reasoning over documents | Weight 5% | Carbon 28% | Behind Muse Wonder Full, 33% |
The surprise in the knowledge test
- Accuracy: Claude Fable 5.1 68%, Claude Opus 5.5 66%, GPT-6.1 Sol and GPT-6 Astra 62%, Carbon 58%
- Hallucination rate: Carbon 11%, Claude Opus 5.5 and GPT-6 Astra about 50%
- Carbon wins because it rarely bluffs, not because it knows the most
Read the weights
- Each of the four blocks counts 25%: knowledge work, coding, science, general
- Carbon trails in two of the four heaviest tests (15% each): office-style work and real-world code
- Hallucination rate = wrong answers among all questions not answered correctly
Source: Artificial Analysis, 28 Oct 2026; Index v5 announcement, 27 Oct 2026; Artificial Analysis model data, retrieved 29 Oct 2026. Scores rounded as in the article; Carbon tested at high, other models at their highest tested setting.
Powered by Thorsten Meyer AI
A Narrow Lead, Not a Universal Win
The one-point margin makes the headline ranking easy to overread. The index combines results across selected tasks, and the leader changes by category. A company choosing a model for coding, office work or document analysis may care more about that specific result than the overall score. The benchmark therefore offers a comparison, not a direct answer about which model is best for every user.
Price and access also affect the practical value of the score. The reported per-task estimate puts Carbon below Opus 5.5 but above two other named models, while public access has not begun. For businesses, the results can help identify models to test, but performance and cost on their own workloads remain unestablished by this leaderboard. For everyday users, the ranking does not mean Carbon is available to use now.
Products Worth Considering
Apple MacBook Pro 16.2-inch with M5 Pro, 18-core CPU, 20-core GPU, 48GB Memory, 1TB SSD, Wi-Fi 7, Space Black
Apple MacBook Pro 16-inch (2021) – M1 Pro Chip, 10-Core CPU, 16-Core GPU, 16GB RAM, 1TB SSD, Space Gray
Logitech MX Keypad Programmable Macro Pad, Graphite
As an affiliate, we earn on qualifying purchases.
How Index v5 Builds Its Score
Artificial Analysis describes the index as a combined score across nine tests in four equally weighted blocks: knowledge work, coding, science and general capability. The source report says 60% of version 5 uses private test sets or held-out answers, and many tasks involve longer agent workflows such as working in a terminal, producing reports or fixing code. These choices shape what the score rewards and how much it costs to generate.
The report highlights a distinction in AA-Omniscience: Carbon’s score is not based on having the highest plain accuracy. It says Carbon answered 58% correctly, while several Claude and GPT models answered more correctly, but Carbon scored higher on a measure that penalizes incorrect answers. The report gives Carbon an 11% hallucination rate on that test, compared with about 50% for Claude Opus 5.5 and GPT-6 Astra. In this test, the result rewards restraint as well as correct answers; it should not be read as proof that Carbon knows more overall.
The leaderboard’s prior version was rebuilt shortly before Carbon took the top spot. The source says Claude Opus 5.5 led the Tuesday ranking before Google’s model moved into first place the following day. That timing makes the result a fresh update to a changing benchmark, rather than a settled assessment across every evaluation method.
Access and Benchmark Limits
Several details remain open. Google has not provided a public release date in the source material, which says Carbon is being offered first to “a subset of cyber defenders”, then to trusted partners and later to developers, businesses and consumers. The timing and terms of each stage are not specified.
The source is a report summarizing Artificial Analysis results; it does not include the full benchmark data or independent replication in the material provided. The index measures performance on its nine selected tests and particular model settings. Results may differ under other tasks, prompts, tools or evaluation methods. The available information also does not establish how Carbon will perform on individual organizations’ workloads or what their final costs will be.
Google’s Staged Rollout
According to the source report, Google plans to move from a limited group of cyber defenders to trusted partners, then to developers, businesses and consumers. No dates for those steps are given. Until access expands, users and organizations cannot verify the leaderboard result against their own work through ordinary public use.
For now, the next useful evidence will be broader availability, additional task-level comparisons and real-world cost and performance reports. The index can guide which models merit testing, but the one-point difference alone does not settle which system is the right fit for a particular task.
Key Questions
What is Gemini 4 Carbon’s score?
It scored 59 points on Artificial Analysis Intelligence Index v5, according to the source report.
How far ahead is Carbon?
Carbon is one point ahead of Claude Opus 5.5 at 58 and two points ahead of GPT-6 Astra at 57 in the reported ranking.
Is Gemini 4 Carbon available to the public?
Not yet, according to the report. Google is starting with a subset of cyber defenders and has not announced dates for wider access.
Does the top ranking mean Carbon is best at every task?
No. The report says Carbon placed first in four of nine tests and lost to other models in several categories, including office-style knowledge work and real-world coding.
How much does Carbon cost to run?
The report estimates $7.76 per index task at standard prices and lists introductory API rates of $2 per million input tokens and $10 per million output tokens. Costs for a specific use case may vary.
Source: ThorstenMeyerAI.com
