Llama 3 8B Instruct
Meta developed and released the Meta Llama 3 family of large language models (LLMs), a collection of pretrained and instruction tuned generative text models in 8 and 70B sizes.
text → text · made by Meta
Sold by 6 ways
| Seller | Lane | Rate |
|---|---|---|
| DigitalOcean cloud | standard | $0.2per Mtok in$0.2per Mtok outraw.githubusercontent.com · read 2026-10-31 |
| AWS Bedrock cloud | standard | $0.3per Mtok in$0.6per Mtok outraw.githubusercontent.com · read 2026-10-31 |
| DeepInfra aggregator | standard | $0.03per Mtok in$0.04per Mtok outapi.deepinfra.com · read 2026-08-25 |
| Novita AI 0 ways | $0.04per Mtok in$0.04per Mtok out | |
| Novita AI aggregatornot on the seller's list today | standard | $0.04per Mtok in$0.04per Mtok outapi.novita.ai · read 2026-08-25 |
| Together AI aggregator | standard | $0.2per Mtok in$0.2per Mtok outraw.githubusercontent.com · read 2026-10-31 |
| Featherless AI aggregator | standard | $0.37per Mtok in$0.38per Mtok outapi.featherless.ai · read 2026-10-29 |
Measured 11 standings
| Place | Board | Metric | Score |
|---|---|---|---|
| 3rdof 15 | Adversarial nli — Epoch AI | Score | 57.3 |
| 9thof 70 | Open book qa — Epoch AI | Accuracy | 82.6 |
| 17thof 134 | Arc ai2 — Epoch AI | Challenge score | 0.828 |
| 78thof 115 | Trivia qa — Epoch AI | EM | 0.677 |
| 80thof 92 | Forecastbench — Epoch AI | Overall score | 59.4 |
| 80thof 217 | Mmlu — Epoch AI | EM | 0.688 |
| 104thof 108 | Math level 5 — Epoch AI | mean_score | 6.127 |
| 114thof 137 | Wino grande — Epoch AI | Accuracy | 65 |
| 232ndof 235 | Dtbench — Epoch AI | Accuracy | 43.93 |
| 290thof 306 | OTIS Mock AIME 2024-2025 | mean_score | 1.944 |
| 315thof 326 | GPQA Diamond | mean_score | 26.073 |
About
Llama 3 8B Instruct — a text model from Meta, sold by 5 companies from $0.03 in and $0.04 out per million tokens, placed 9th of 70 on Open book qa — Epoch AI.
It takes text and returns text, with a context window of 32,768 tokens. It was published in April 2024, trained on material up to December 2023. Its sellers say it can call a tool. The catalogue files it under reasoning and agents. Five companies sell it. The cheapest is $0.03 in and $0.04 out per million tokens at DeepInfra. It has been measured on 11 boards, and stands best at 9th of 70 on Open book qa — Epoch AI.
Every current figure
- Maker
- Meta
- Register
- model
- Takes
- text
- Returns
- text
- Context
- 32,768 tokens
- Published
- April 2024
- Knowledge to
- December 2023
- Parameters
- 8 billion · read from its own name
- Sellers
- 5
- Maker's own price
- not read
- Price
- $0.03 in and $0.04 out per million tokens — DeepInfra
- Boards
- 11
- Best place
- 9th of 70 — Open book qa — Epoch AI
Known as 14 names
Meta-Llama-3-8B-Instructmeta-llama/Meta-Llama-3-8B-InstructAI-Sweden-Models/Llama-3-8B-instructByteResearch/Llama-3-8B-InstructISTA-DASLab/Meta-Llama-3-8B-InstructKJCHUA/Meta-Llama-3-8B-InstructMaziyarPanahi/Llama-3-8B-Instruct-64kNousResearch/Meta-Llama-3-8B-InstructPawanKrd/Meta-Llama-3-8B-Instructgradientai/Llama-3-8B-Instruct-262kkuotient/Meta-Llama-3-8B-Instructtuneai/Meta-Llama-3-8B-Instructunsloth/llama-3-8b-Instructmeta-llama/llama-3-8b-instruct