Side-by-side benchmark comparison across knowledge, coding, math, and reasoning.
GPT-5.2-Codex wins overall with a score of 82 vs 48 (34 point difference).GPT-5.2-Codex wins 4 out of 4 categories.
GPT-5.2-Codex
96
Llama 3 70B
56.5
GPT-5.2-Codex
95
Llama 3 70B
50
GPT-5.2-Codex
97.1
Llama 3 70B
57
GPT-5.2-Codex
94
Llama 3 70B
55
GPT-5.2-Codex scores higher overall with 82 vs 48, a difference of 34 points across all benchmarks.
GPT-5.2-Codex leads in knowledge tasks with an average score of 96 vs 56.5.
GPT-5.2-Codex leads in coding with an average score of 95 vs 50.
GPT-5.2-Codex leads in math with an average score of 97.1 vs 57.
GPT-5.2-Codex leads in reasoning with an average score of 94 vs 55.