Claude Fable 5.2 moves above Gemini 4 Carbon on a top AI index

This week, the model Claude Fable 5.2 of Anthropic moved above Gemini 4 Carbon of Google on the Artificial Analysis Intelligence Index. Each company uses this scorecard, and it is not from them. Meta released a model at $0.10 for each million input tokens. OpenAI gives its top model for computer security only to users after a check.

How it works

Meta made its model for the basic work. A top model makes the plan. The source gives no data on the check that OpenAI does.

The numbers

• Meta price: $0.10 for each million input tokens.
• Fable 5.2 price: $8 for each million input tokens, down from $10, Anthropic tells.
• Index scores: none given.

Anthropic gave no test scores for coding, reasoning or vision.

What is not known

The scores of the 2 models are not known. The name and the test results of the Meta model are not known. The instructions of the OpenAI check are not known.

Insight: Top rank and low price are sold apart

Anthropic tells that Fable 5.2 has a price of $8 for each million input tokens. The Meta model has a price of $0.10, which is 80 times less. It does the basic work with a model that makes the plan. This reads as a market with 2 parts: top models that make plans and models with a low price that work. Possibly the top model holds the customer. Select which tasks must have the top model.