Claude Instant 1.2 — benchmark results

Anthropic Claude Instant 1.2 legacy lower-latency chat model. Provider: Anthropic. Released 2023-08-09. Access: API.

Unified ELO 1468 ± 26, rank #918 of 1776 rated models, from 13 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
ARC Challenge (AI2)86.3Accuracy (%)88.5
GSM8K86.7Accuracy (%)86.7
BenchBench64.86Aggregate Score (%)68.4
MMLU73.2Accuracy (%)63
AlpacaEval 2.025.61LC Win Rate (%)61.7
HELM NaturalQuestions (Open)73.1F1 (%)61.1
HELM WMT 201419.43BLEU-4 (%)55.6
TriviaQA78.7Accuracy (%)51.3
HELM Lite44.55Mean win rate (self-reported)44.2
HELM (Stanford)39.94Mean Win Rate (%)37.8
HELM NaturalQuestions (Closed)34.32F1 (%)35
Epoch AI - ECI121.08ECI Score18.3

Interactive version: theaggregate.ai/model?slug=claude-instant-1-2 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.