Llama 3.3 70B

Proven 70B model with excellent instruction-following

Llama 3.3 70B is Meta's refined 70-billion parameter model that punches above its weight class. It delivers near-GPT-4 performance on instruction-following tasks and is widely used as a cost-effective production model.

Try Llama 3.3 70B

Specifications

Context window128k tokens

Input price$0.100 / 1M tokens

Output price$0.300 / 1M tokens

Credits per query3 cr

Released2024-12

LicenseLlama 3.3 Community

Features

Tool useJSON modeFunction callingCode

Benchmark Scores

What do these mean?

Sourced from official model cards and academic papers. Higher is better.

GPQA Diamond

PhD-level science & reasoning

—

AIME 2024

Mathematical olympiad problems

—

Arena Elo

Chatbot Arena human preference ranking

—

HumanEval

Python coding — pass@1 rate

88.4%

MATH

Competition-level mathematics

77%

MMLU

General knowledge across 57 academic domains

86%

MT-Bench

Multi-turn instruction following (out of 10)

—

SWE-Bench

Real-world GitHub issue resolution rate

—

Run locally

Download and run this model on your own hardware — no API key or internet required after setup.

Min VRAM40 GB

Rec. VRAM42 GB

Best VRAM75 GB

Parameters70B

Speed~4 tok/s

Context128K tokens

Ollama

Terminal

Run this command in your terminal. Ollama downloads and manages the model locally.

ollama pull llama3.3:70b

LM Studio

Desktop app · no terminal needed

Hugging Face

Raw weights · any runtime

More from Meta

Llama 4 Behemoth3 cr

Meta's most powerful open-weight model

Llama 4 Maverick3 cr

Meta's frontier multimodal model

Llama 4 Scout3 cr

10M-token context with MoE efficiency