🔍 Read the full analysis: What Changes When You Use Claude Fable 5.2 Or Gemini 4 At Work? on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Anthropic’s Claude Fable 5.2 scored higher than Google’s Gemini 4 models on the Artificial Analysis Intelligence Index, but Gemini 4 Carbon led several individual work tests and cost less per index task. The tests show no single winner: results vary by task, model setting and price, while Fable answered more questions accurately but guessed incorrectly more often when it lacked an answer.
Claude Fable 5.2 leads Google’s new Gemini 4 models on a broad independent AI index, but a November 17 comparison of published tests finds Gemini 4 Carbon ahead on several workplace tasks and cheaper per index task. The results suggest that which model fits a team depends on its specific work, tolerance for errors and budget—not just the overall ranking.
On the Artificial Analysis Intelligence Index, Claude Fable 5.2 scored 65 at its Max effort setting and 62 at High. Gemini 4 Carbon scored 60 at High, Gemini 4 Hydrogen scored 54 at its highest setting, and Gemini 4 Argon scored 50 at High. These scores summarize a range of tests; individual task results do not all follow the same order.
The comparison reports that Fable 5.2 High led Carbon on coding in open-source repositories, with scores of 49.1% and 40.6%, respectively, and on long-document reasoning, at 83.3% versus 79.3%. Carbon led on terminal tasks, professional-document reasoning and business workflow automation. On the professional-document test, both models scored below 30%: Carbon received 28.4% and Fable 21.8%.
Estimated cost for one Artificial Analysis index task was $26.54 for Fable at Max and $9.08 at High, compared with $5.48 for Carbon, $4.97 for Argon and $3.16 for Hydrogen. Carbon and Argon estimates use Google’s introductory prices, which were scheduled to double on December 1. These are task-level estimates, not a guarantee of what an individual company will pay; token use and model settings affect the bill.
How Task Choice Changes the Cost
For companies deciding which model to use, the comparison makes the trade-off concrete: a higher overall score does not mean the model leads every workflow. Teams focused on repository coding or long documents may favor Fable 5.2 based on these results, while those automating business software or handling terminal tasks may find Carbon’s test performance more relevant. Benchmark scores are evidence to guide trials, not proof that a model will perform the same way on a company’s own systems and data.
Price also changes the calculation. Carbon and Hydrogen had lower reported index-task costs than Fable 5.2 at either High or Max, although Google’s Carbon and Argon introductory prices were due to expire on December 1. A lower cost per benchmark task may not translate directly to lower spending in production: real workloads differ in prompt length, output, tool use and reasoning effort. Buyers need to compare performance and cost on representative tasks before switching or standardizing.
Reliability presents another trade-off. In the reported evaluation, Fable answered more questions correctly, but it was also more likely to give a wrong answer when it did not know one. For workplaces where a confident error could affect customer service, research or operational decisions, how a model handles uncertainty may matter as much as its accuracy on questions it can answer.
Products Worth Considering

Apple Mac mini Desktop Computer with M6 Chip (2026)

Canon SELPHY CP1500 Compact Photo Printer Bundle with Two KP-108IN Ink and Paper Sets, Black

Apple Mac mini M6, 12-core CPU and GPU, 24GB Memory, 1TB SSD
As an affiliate, we earn on qualifying purchases.
Three Gemini Models, Different Roles
Anthropic released Claude Fable 5.2 on November 9, describing it as a model for demanding, open-ended technical work. Google made Gemini 4 Carbon available to paying customers on November 10, then released Gemini 4 Hydrogen to all users, including through Google Search, on November 12. Gemini 4 is a model family rather than a single system: Google calls Carbon its most capable reasoning model and recommends Argon for most work, while Hydrogen is positioned for faster agentic workflows.
The models also differ in input support and pricing. The source comparison says all three Gemini 4 models and Fable 5.2 have a one-million-token context window. Carbon and Hydrogen accept text, images, speech and video, according to Artificial Analysis; Argon and Fable 5.2 accept text and images. Anthropic’s listed price for Fable 5.2 is $8 per million input tokens and $40 per million output tokens. Google listed Carbon and Argon at $2 and $10 through November 30, rising to $4 and $20 on December 1; Hydrogen was listed at $1.50 and $7.50.
The comparison’s author disclosed that the article was researched and drafted with AI assistance, including Anthropic’s Claude, and that an Anthropic model was one of the models being compared. The author said Anthropic, Google and Artificial Analysis did not review or sponsor the article. Its benchmark figures were accessed November 17, 2026.
““the most ambitious, open-ended technical work””
— Anthropic
Benchmark Results Leave Open Questions
The figures are benchmark results, not workplace guarantees. The source comparison does not provide enough detail here to establish how the tested tasks match any particular company’s workload, or how performance changes with different prompts, tools and data. The results also vary by test, so the overall index cannot resolve which model is best for every role.
The comparison reports a marked difference in answers the models did not know: Fable 5.2 had 77.2% accuracy against Carbon’s 58.3%, but its reported hallucination rate was 58.0%, compared with 11.4% for Carbon. The definitions and evaluation details behind those reliability figures are not set out in the supplied material, so they should be read as results from that evaluation rather than universal rates of error.
Pricing estimates may change after Google’s introductory rates end, and per-task costs depend on how much each model uses. It is also unclear from the comparison how the models would perform on a given organization’s private documents or software. Anthropic recommends starting with Opus 5.5 for most work, while Google recommends Argon for most software engineering and knowledge work; those vendor recommendations are distinct from the independent test results.
Test Before Choosing a Model
The next practical step for a workplace is to test candidate models on its own representative tasks, using the same prompts, tools and success criteria. Teams can compare answer quality, error handling, task completion time and total token costs, then decide whether one model should handle all work or different models should be assigned to different jobs.
Companies weighing Carbon or Argon pricing also need to account for Google’s scheduled December 1 price increase. Further independent evaluations could clarify how the models perform across additional tasks and whether their benchmark strengths persist in production. The source material does not give a later release or evaluation date.
Key Questions
Which model scored highest overall?
Claude Fable 5.2 at Max scored 65 on the Artificial Analysis Intelligence Index, ahead of Gemini 4 Carbon at 60. Fable scored 62 at High. The ranking does not mean it led every individual test.
Is Gemini 4 better for workplace tasks?
It depends on the task. In the cited tests, Carbon led on terminal work, professional documents and business workflows, while Fable 5.2 led on open-source repository coding and long-document reasoning.
Which model had the lowest reported cost per index task?
Gemini 4 Hydrogen had the lowest reported estimate, at $3.16 per index task. Fable 5.2 cost $9.08 at High and $26.54 at Max. These are benchmark-based estimates, and actual costs depend on usage and model settings.
When were Google’s Carbon and Argon prices scheduled to change?
The introductory prices cited for Carbon and Argon were scheduled to double on December 1, 2026, from $2 per million input tokens and $10 per million output tokens to $4 and $20, respectively.
Source: ThorstenMeyerAI.com
