Compare Llama 2 Chat 70B vs. Llama 3 70B Instruct

Pricing About Us Blog Docs

Book a demo

Compare

Understand and compare Llama 2 Chat 70B vs. Llama 3 70B Instruct

Try Podial

Turn your documents into engaging podcast discussions.

Overview

Llama 2 Chat 70B was released 9 months before Llama 3 70B Instruct.

	Llama 2 Chat 70B	Llama 3 70B Instruct
Provider The entity that provides this model.	Meta	Meta
Input Context Window The number of tokens supported by the input context window.	4,096 tokens	8,000 tokens
Maximum Output Tokens The number of tokens that can be generated by the model in a single request.	2,048 tokens	2,048 tokens
Release Date When the model was first released.	July 18th, 2023 over 1 year ago 2023-07-18	April 18th, 2024 7 months ago 2024-04-18
Knowledge Cutoff Limit on the knowledge base used by the model.	September 2022 September 2022	December 2023 December 2023
Open Source	Yes	Yes
API Providers The providers that offer this model. (This is not an exhaustive list.)	Azure AI, AWS Bedrock, Google Cloud Vertex AI Model Garden, NVIDIA NIM, IBM watsonx	Azure AI, AWS Bedrock, Google Cloud Vertex AI Model Garden, NVIDIA NIM, IBM watsonx

Pricing

	Llama 2 Chat 70B	Llama 3 70B Instruct
Input Cost of input data provided to the model.	Pricing not available.	Pricing not available.
Output Cost of output tokens generated by the model.	Pricing not available.	Pricing not available.

Benchmarks

Compare relevant benchmarks between Llama 2 Chat 70B and Llama 3 70B Instruct.

	Llama 2 Chat 70B	Llama 3 70B Instruct
MMLU Evaluating LLM knowledge acquisition in zero-shot and few-shot settings.	68.9 Source	82.0 (5-shot) Source
MMMU A wide ranging multi-discipline and multimodal benchmark.	30.1 Source	Benchmark not available.
HellaSwag A challenging sentence completion benchmark.	85.3 (0-shot) Source	Benchmark not available.
GSM8K Grade-school math problems benchmark.	56.8 (8-shot) Source	93.0 (8-shot) Source
HumanEval A benchmark to measure functional correctness for synthesizing programs from docstrings.	29.9 (0-shot) Source	Benchmark not available.
MATH Benchmark performance on Math problems ranging across 5 levels of difficulty and 7 sub-disciplines.	Benchmark not available.	51.0 (0-shot) Source

Llama 2 Chat 70B, developed by Meta, features a context window of 4096 tokens. The model was released on July 18, 2023, and has achieved a score of 30.1 in the MMMU benchmark and 68.9 in the MMLU benchmark.

Llama 3 70B Instruct, developed by Meta, features a context window of 8000 tokens. The model was released on April 18, 2024, and achieved a score of 82.0 in the MMLU benchmark under a 5-shot scenario.

Compare more models

Llama 2 Chat 70B vs.

Llama 3.1 70B Instruct

Llama 2 Chat 70B vs.

GPT-3.5 Turbo 0125