Claude Max News

YES

YES

4% to under 6% (3–4 solved)

9% or higher (7+ solved)

Under 4% (no model beats GPT-6 Astra's 2 of 68)

Google

Anthropic

OpenAI

Meta

xAI

Gemini 4 Pro 5-8 points worse

Gemini 4 Pro 0-4 points worse

Gemini 4 Pro ≥9 points worse

NO

2029

2030

2031

2027

2028

2032

2033

2034

2035

2026

46% or higher

Below 38%

38% to below 42%

NO

NO

Below 78%

78% to below 82%

82% to below 88%