Back to Models

Llama 3.3 70B Instruct

active
Meta2024-12-06
Specifications
Modalities
text
Context Window128K tokens
Max Output-
Training Cutoff2023-12-01
Open SourceYes
Pricing (per 1M tokens)
Input$-
Output$-
Benchmarks
Arena score (text)1318

Arena ratings: LMArena leaderboard dataset, CC BY 4.0, snapshot 2026-10-28.

MMLU-Pro68.9
SWE-Bench-
GPQA Diamond50.5
HumanEval88.4
ARC-AGI-
MATH77.0
API
Model IDmeta-llama/Llama-3.3-70B-Instruct
Sources