Claude 2

Name: Claude 2
Author: Anthropic

Anthropic · 2023-07-11

Claude 2 is Anthropic's second-generation Claude model, improving over Claude 1 with longer responses, a larger context window, and stronger coding, math, and reasoning.

Benchmark scores

Benchmark	Score
AlpacaEval 2.0 (length-controlled win rate)	28.2
AlpacaEval 2.0 (raw win rate)	17.2
Artificial Analysis Coding Index	12.9
Artificial Analysis Progress Index v1	128
Capability	110.2
DTBench	51.9
Epoch Capabilities Index (ECI)	120
EQ-Bench v2	72.9
GPQA Diamond	34.7
GPQA Diamond (Artificial Analysis avg@5)	34.4
LiveCodeBench (Artificial Analysis)	17.1
MATH Level 5	11.7
MMLU Pro	48.6
OTIS Mock AIME 2024-2025 (Epoch legacy model grader)	2.5
SciCode (Artificial Analysis avg@3)	19.4