Model Family · Meta

Llama

Name: Llama model family
Creator: Meta

Llama is Meta’s open-weight language model family. Since the original 2023 release it has become the most widely fine-tuned foundation model in the world, with variants spanning 1B to 405B parameters across dense and Mixture-of-Experts architectures.

See all models from Meta

Models in family

Open weight

API only

Avg score

50.0

Top benchmark

85.0

AIME 2026

Total HF downloads

14.5M

Primary modality

Text

Context window

2K – 10.0M

First release

Feb 2023

Latest release

Oct 2025

Llama Family Leaderboard

Every release in the Llama family, ranked by composite score across benchmarks, popularity, efficiency, and versatility.

#	Model	Modality	Score	Params	Released
1	Llama 3.1 8B Instruct	text	BB69.8	8B	Jul 2024
2	Llama 3 8B Instruct	text	BB63.5	8B	Apr 2024
3	llama-embed-nemotron-8b	embedding	BB61.4	7.5B	Oct 2025
4	Llama 3.3 70B Instruct	text	BB57.3	70B	Dec 2024
5	LLaMA 65B	text	BB55.9	65B	Feb 2023
6	Llama 3.1 70B Instruct	text	CC50.8	70B	Jul 2024
7	Llama 3 70B Instruct	text	CC48.9	70B	Apr 2024
8	Llama 2 7B Chat	text	CC48.6	7B	Jul 2023
9	Llama 4 Scout	text	CC43.5	109B	Apr 2025
10	Llama 2 13B Chat	text	CC40.9	13B	Jul 2023

See All Models

Best in Each Modality

When a family spans multiple modalities, the strongest model in each one.

Text

Llama 3.1 8B Instruct

69.8

Embedding

llama-embed-nemotron-8b

61.4

Modality Breakdown

How the family splits across text, image, video, audio, and embedding releases.

Text12 (92%)

Embedding1 (8%)

Release Timeline

When each release shipped, newest first. Useful for tracking version cadence.

2025
- llama-embed-nemotron-8b
  Oct 21
  61.4
- Llama 4 Scout
  Apr 4
  43.5
- Llama 4 Maverick
  Apr 4
  33.0
2024
- Llama 3.3 70B Instruct
  Dec 5
  57.3
- Llama 3.1 405B Instruct
  Jul 22
  35.8
- Llama 3.1 70B Instruct
  Jul 22
  50.8
- Llama 3.1 8B Instruct
  Jul 22
  69.8
- Llama 3 70B Instruct
  Apr 17
  48.9
- Llama 3 8B Instruct
  Apr 17
  63.5
2023
- Llama 2 70B Chat
  Jul 17
  40.4
- Llama 2 13B Chat
  Jul 17
  40.9
- Llama 2 7B Chat
  Jul 17
  48.6
- LLaMA 65B
  Feb 23
  55.9

Grade Distribution

Composite grades across this family. Higher is better, blending benchmarks, popularity, and efficiency.

AGrade A0 (0%)

BGrade B5 (38%)

CGrade C6 (46%)

DGrade D2 (15%)

FGrade F0 (0%)

Open-Weight Llama Releases

Models with downloadable weights, ranked by composite score.

#	Model	Modality	Score	Params	Released
1	Llama 3.1 8B Instruct	text	BB69.8	8B	Jul 2024
2	Llama 3 8B Instruct	text	BB63.5	8B	Apr 2024
3	llama-embed-nemotron-8b	embedding	BB61.4	7.5B	Oct 2025
4	Llama 3.3 70B Instruct	text	BB57.3	70B	Dec 2024
5	LLaMA 65B	text	BB55.9	65B	Feb 2023
6	Llama 3.1 70B Instruct	text	CC50.8	70B	Jul 2024
7	Llama 3 70B Instruct	text	CC48.9	70B	Apr 2024
8	Llama 2 7B Chat	text	CC48.6	7B	Jul 2023
9	Llama 4 Scout	text	CC43.5	109B	Apr 2025
10	Llama 2 13B Chat	text	CC40.9	13B	Jul 2023
11	Llama 2 70B Chat	text	CC40.4	70B	Jul 2023
12	Llama 3.1 405B Instruct	text	DD35.8	405B	Jul 2024
13	Llama 4 Maverick	text	DD33.0	400B	Apr 2025

Frequently Asked Questions

What is the Llama family?

The Llama family is a series of AI models from Meta. This page lists every release in the family with its benchmark scores, parameter count, and hardware requirements.

Which Llama model should I use?

By composite score, Llama 3.1 8B Instruct is currently the top model in the family. For local inference, match the parameter count to your VRAM budget. For quality, pick the highest scorer that fits.

Are Llama models open source?

See the open-weight section above for models you can run locally. The API-only section lists closed releases that must be accessed through the provider’s API.

Where to Run Llama?

Spin up an instance in the cloud, or pick local hardware that fits.

Full Directory

Compare With Every Other Model

Open the full directory to filter by hardware, capability, license, and benchmark score.

Open Directory

Or Browse by Provider

Look up OpenAI, Anthropic, Meta, and Every Provider

See every model from a lab side by side, with aggregate stats. Useful when you want a cross-family view of one provider.

Browse Providers

Need Help Picking a Family Member?

We help teams match a model size to their hardware, budget, and latency target. Free 20-minute call.

Model Family · Meta

Llama

See all models from Meta

Models in family

Open weight

API only

Avg score

50.0

Top benchmark

85.0

AIME 2026

Total HF downloads

14.5M

Primary modality

Text

Context window

2K – 10.0M

First release

Feb 2023

Latest release

Oct 2025

Llama Family Leaderboard

Every release in the Llama family, ranked by composite score across benchmarks, popularity, efficiency, and versatility.

#	Model	Modality	Score	Params	Released
1	Llama 3.1 8B Instruct	text	BB69.8	8B	Jul 2024
2	Llama 3 8B Instruct	text	BB63.5	8B	Apr 2024
3	llama-embed-nemotron-8b	embedding	BB61.4	7.5B	Oct 2025
4	Llama 3.3 70B Instruct	text	BB57.3	70B	Dec 2024
5	LLaMA 65B	text	BB55.9	65B	Feb 2023
6	Llama 3.1 70B Instruct	text	CC50.8	70B	Jul 2024
7	Llama 3 70B Instruct	text	CC48.9	70B	Apr 2024
8	Llama 2 7B Chat	text	CC48.6	7B	Jul 2023
9	Llama 4 Scout	text	CC43.5	109B	Apr 2025
10	Llama 2 13B Chat	text	CC40.9	13B	Jul 2023

See All Models

Best in Each Modality

When a family spans multiple modalities, the strongest model in each one.

Text

Llama 3.1 8B Instruct

69.8

Embedding

llama-embed-nemotron-8b

61.4

Modality Breakdown

How the family splits across text, image, video, audio, and embedding releases.

Text12 (92%)

Embedding1 (8%)

Release Timeline

When each release shipped, newest first. Useful for tracking version cadence.

2025
- llama-embed-nemotron-8b
  Oct 21
  61.4
- Llama 4 Scout
  Apr 4
  43.5
- Llama 4 Maverick
  Apr 4
  33.0
2024
- Llama 3.3 70B Instruct
  Dec 5
  57.3
- Llama 3.1 405B Instruct
  Jul 22
  35.8
- Llama 3.1 70B Instruct
  Jul 22
  50.8
- Llama 3.1 8B Instruct
  Jul 22
  69.8
- Llama 3 70B Instruct
  Apr 17
  48.9
- Llama 3 8B Instruct
  Apr 17
  63.5
2023
- Llama 2 70B Chat
  Jul 17
  40.4
- Llama 2 13B Chat
  Jul 17
  40.9
- Llama 2 7B Chat
  Jul 17
  48.6
- LLaMA 65B
  Feb 23
  55.9

Grade Distribution

Composite grades across this family. Higher is better, blending benchmarks, popularity, and efficiency.

AGrade A0 (0%)

BGrade B5 (38%)

CGrade C6 (46%)

DGrade D2 (15%)

FGrade F0 (0%)

Open-Weight Llama Releases

Models with downloadable weights, ranked by composite score.

#	Model	Modality	Score	Params	Released
1	Llama 3.1 8B Instruct	text	BB69.8	8B	Jul 2024
2	Llama 3 8B Instruct	text	BB63.5	8B	Apr 2024
3	llama-embed-nemotron-8b	embedding	BB61.4	7.5B	Oct 2025
4	Llama 3.3 70B Instruct	text	BB57.3	70B	Dec 2024
5	LLaMA 65B	text	BB55.9	65B	Feb 2023
6	Llama 3.1 70B Instruct	text	CC50.8	70B	Jul 2024
7	Llama 3 70B Instruct	text	CC48.9	70B	Apr 2024
8	Llama 2 7B Chat	text	CC48.6	7B	Jul 2023
9	Llama 4 Scout	text	CC43.5	109B	Apr 2025
10	Llama 2 13B Chat	text	CC40.9	13B	Jul 2023
11	Llama 2 70B Chat	text	CC40.4	70B	Jul 2023
12	Llama 3.1 405B Instruct	text	DD35.8	405B	Jul 2024
13	Llama 4 Maverick	text	DD33.0	400B	Apr 2025

Frequently Asked Questions

What is the Llama family?

The Llama family is a series of AI models from Meta. This page lists every release in the family with its benchmark scores, parameter count, and hardware requirements.

Which Llama model should I use?

By composite score, Llama 3.1 8B Instruct is currently the top model in the family. For local inference, match the parameter count to your VRAM budget. For quality, pick the highest scorer that fits.

Are Llama models open source?

See the open-weight section above for models you can run locally. The API-only section lists closed releases that must be accessed through the provider’s API.

Where to Run Llama?

Spin up an instance in the cloud, or pick local hardware that fits.

Full Directory

Compare With Every Other Model

Open the full directory to filter by hardware, capability, license, and benchmark score.

Open Directory

Or Browse by Provider

Look up OpenAI, Anthropic, Meta, and Every Provider

See every model from a lab side by side, with aggregate stats. Useful when you want a cross-family view of one provider.

Browse Providers

Need Help Picking a Family Member?

We help teams match a model size to their hardware, budget, and latency target. Free 20-minute call.

Llama

Quick Facts

Llama Family Leaderboard

Best in Each Modality

Modality Breakdown

Release Timeline

2025

2024

2023

Grade Distribution

Open-Weight Llama Releases

Frequently Asked Questions

What is the Llama family?

Which Llama model should I use?

Are Llama models open source?

Where to Run Llama?

Compare With Every Other Model

Look up OpenAI, Anthropic, Meta, and Every Provider

Need Help Picking a Family Member?

Llama

Quick Facts

Llama Family Leaderboard

Best in Each Modality

Modality Breakdown

Release Timeline

2025

2024

2023

Grade Distribution

Open-Weight Llama Releases

Frequently Asked Questions

What is the Llama family?

Which Llama model should I use?

Are Llama models open source?

Where to Run Llama?

Compare With Every Other Model

Look up OpenAI, Anthropic, Meta, and Every Provider

Need Help Picking a Family Member?