Does GPT-6 Astra actually consume fewer tokens than Claude Fable 5.1 and older models for the same tasks?

Recent token-usage comparisons indicate that GPT-6 Astra consumes significantly fewer tokens than Claude Fable 5.1 and older models for similar tasks. Artificial Analysis data suggests Astra uses approximately 21k output tokens per task, while Claude Fable 5.1 uses around 64k. Additionally, claims suggest Astra uses about 1/3 the tokens of GPT-5.6 Sol and 1/5 the tokens of Claude Opus 5 in certain coding-agent benchmarks, highlighting its efficiency.

This report cites specific Artificial Analysis data, showing GPT-6 Astra using 21k output tokens per task compared to Claude Fable 5.1's 64k, unlike general claims.