NEWHow AI models draw the world map · Gecko Tests
Beta
Home/Models/Llama 3 70B Instruct
Meta logo

Llama 3 70B Instruct

by Meta · Released Apr 2024

Open Source
30.6
avg score
Rank #244
Compare
Better than 22% of all models
Context
8K tokens (~4 books)
Input $/1M
$0.51
Output $/1M
$0.74
Type
text
License
Open Source
Benchmarks
9 tested
Data as of
About

Meta's latest class of model (Llama 3) launched with a variety of sizes & flavors. This 70B instruct-tuned version was optimized for high quality dialogue usecases. It has demonstrated strong...

Tested on 9 benchmarks · BenchGecko score 30.6. Top scores: Chatbot Arena Elo — Overall (1275.1%), MMLU (72.4%), Winogrande (67.0%).

Looking for similar performance at lower cost?
Llama 3.1 8B Instruct scores 30.7 (100% as good) at $0.02/1M input · 96% cheaper
Capabilities
coding
5.0
#195 globally
math
13.4
#249 globally
knowledge
45.4
#161 globally
Benchmark Scores
Compare All
Tested on 9 benchmarks · Ranked across 4 categories
Score Distribution (all 312 models)
0255075100
▲ You are here
codingCompare coding →
Cybench

Capture-the-flag cybersecurity challenges. Tests vulnerability analysis, reverse engineering, cryptography, and exploitation skills.

5.0·
mathCompare math →
MATH level 5

Competition-level math from AMC, AIME, and olympiad problems. Level 5 is the hardest tier, requiring creative problem-solving.

22.6·
OTIS Mock AIME 2024-2025

Mock AIME (American Invitational Mathematics Exam) problems from OTIS. Tests mathematical competition performance.

4.2·
knowledgeCompare knowledge →
MMLU

Massive Multitask Language Understanding. 57 subjects from STEM, humanities, and social sciences. The most widely-cited knowledge benchmark.

72.4·
Winogrande

Commonsense coreference resolution. Tests understanding of pronoun references in ambiguous sentences.

67.0·
CMMLU

Chinese MMLU. Comprehensive knowledge test specifically designed for Chinese language and culture.

36.9·
Quick compare:Meta logovs Meta Llama 3 8B InstructMeta logovs Llama 3.1 8B Instructvs StarCoder 2 15B
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Model Family · Meta Llama 3
Apr 2024
30.6
$0.51/M in8Kctx9 benchmarks
Apr 2024
41.8+11.2
$0.14/M in(-0.37)8Kctx16 benchmarks
Jan 2024
41.6-0.2
n/an/actx18 benchmarks
See the full Llama 3 family →
Similar Models
Meta logo
Meta Llama 3 8B Instruct
Meta
30.7n/a
Meta logo
Llama 3.1 8B Instruct
Meta
30.7$0.02/1M
StarCoder 2 15B
Unknown
30.5n/a
Links
Info
MetaPricing explorerDevelopers · API
Research
Technical Report
Documentation
API DocsPlayground
Community
@Meta
Source Code
GitHubHugging Face
BenchGecko API
llama-3-70b-instruct
Specifications
  • Typetext
  • Context8K tokens (~4 books)
  • ReleasedApr 2024
  • LicenseOpen Source
  • StatusActive
  • Cost / Message~$0.002
Available On
Meta logoMeta$0.51
Categories
codingmathknowledge
Learn More
transformeropen weightstokens
Share & Export
Tweet
Related Models
Meta logoMeta Llama 3 8B Instruct
30.7
Meta logoLlama 3.1 8B Instruct
30.7
StarCoder 2 15B
30.5
Anthropic logoClaude 3 Sonnet
30.3
Llama 3 70B Instruct is an open-source text AI model by Meta, released in April 2024. It has an average benchmark score of 30.6. Context window: 8K tokens.

Related Models

Meta Llama 3 8B Instruct · MetaLlama 3.1 8B Instruct · MetaStarCoder 2 15B · UnknownClaude 3 Sonnet · Anthropicvicuna-13b-v1.1 · Unknown

Benchmarks

Chatbot Arena Elo — OverallMMLUWinograndeCMMLUOpenBookQA

Related Pages

Meta · ProviderMeta · EconomyAll ModelsCompare ModelsPricingDevelopers · API

Key facts · as of 2026-05-03

  • Llama 3 70B Instruct by Meta. BenchGecko score 30.6, rank 244 of 312 scored models (normalized average of public benchmark scores).
  • List price $0.51 input · $0.74 output per 1M tokens (as of 2026-05-03).

How to cite · data as of 2026-05-03

Llama 3 70B Instruct · benchmarks, pricing and providers. BenchGecko, data as of 2026-05-03. https://benchgecko.ai/model/llama-3-70b-instruct

Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP