Claude Fable 5.2 scores 65 on Artificial Analysis — highest intelligence rating yet

What you could do with it

• A researcher could feed massive documents into the model thanks to its one-million-token context window, roughly equal to 1,500 A4 pages ¹.
• A developer working on tough coding or math challenges could rely on the model to reason through multi-step problems before giving an answer ¹.
• A business user dealing with reading comprehension or automation-heavy workflows could get sharper answers on harder tasks compared with average models ¹.

Top spot, but not the cheapest

Claude Fable 5.2 (Max, Default Fallback) earned an Intelligence Index score of 65, placing it at rank 1 out of 74 models tracked by Artificial Analysis ¹. That is well above the median score of 40 across similar models. For comparison, the index produced 275 million output tokens when evaluating this model, compared with the group median of just 72.8 million — meaning the model writes considerably more text to reach its answers .

The test used nine different evaluations, including coding challenges, automation benchmarks, reading tasks and hard-subset reasoning questions ¹. Anthropic did not publish the results; Artificial Analysis ran them independently on its own infrastructure .

Price and speed

The model carries premium pricing. Input tokens cost $8.00 per million and output tokens cost $40.00 per million, both expensive relative to the median of $2.00 and $10.00 respectively ¹. With a cache discount of 98 percent, the effective cost per intelligence index task drops to around $26.54 — still higher than many competing systems .

Speed is another trade-off. Each task took 44.6 minutes on average, nearly four times slower than the median of 12.7 minutes for comparable models ¹. Users should expect longer waits when asking complex questions.

Built for deep thinking

Claude Fable 5.2 is a reasoning model, meaning it is designed to work through steps before producing a final answer ¹. It supports both text and image input and can handle a context window of one million tokens — roughly 1,500 A4 pages of standard font — making it suitable for analyzing very large documents or long conversations . No public API was listed at the time of reporting .

What to watch next

As other developers release their own reasoning models, the leaderboard may shift. For now, Claude Fable 5.2 holds the top intelligence score, but anyone considering it for daily use will need to weigh accuracy against cost and response time ¹.

How good is it

Claude Fable 5.2 scored 65 out of 100 on the Artificial Analysis Intelligence Index, placing it at rank 1 out of 74 models tracked there ¹. That sits well above the group median of 40, meaning it handles a broad mix of coding, math, reading and reasoning tasks better than most rivals on the list . Artificial Analysis ran the evaluation independently on its own infrastructure across nine tests . The test used 275 million output tokens for this model versus a median of just 72.8 million, showing the win comes from detailed, thorough reasoning rather than concise shortcuts .

Key facts

• Claude Fable 5.2 scored 65 on the Artificial Analysis Intelligence Index — highest so far
• Rank 1 out of 74 tracked models, above the median score of 40
• Pricing: $8/M input tokens and $40/M output tokens, with a 98% cache discount bringing task cost to $26.54
• Reasoning model supporting text and image input with a 1M token context window
• No publicly available API at time of reporting

Sources

1. Claude Fable 5.2 (max) - Intelligence, Performance & Price Analysis | Artificial Analysis