- Services
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 MetaModels in family
14
Open weight
14
API only
0
Avg score
48.9
Top benchmark
88.9
MATH-500
Total HF downloads
5.5M
Primary modality
Text
Context window
2K – 10.0M
First release
Feb 2023
Latest release
Mar 2026
Every release in the Llama family, ranked by composite score across benchmarks, popularity, efficiency, and versatility.
| # | Model | Modality | Score | Params | Released |
|---|---|---|---|---|---|
| 1 | embedding | BB63.8 | 7.5B | Oct 2025 | |
| 2 |
When a family spans multiple modalities, the strongest model in each one.
How the family splits across text, image, video, audio, embedding, and decision releases.
When each release shipped, newest first. Useful for tracking version cadence.
Models with downloadable weights, ranked by composite score.
| # | Model | Modality | Score | Params | Released |
|---|---|---|---|---|---|
| 1 | embedding | BB63.8 | 7.5B | Oct 2025 | |
| 2 |
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.
By composite score, llama-embed-nemotron-8b 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.
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.
Spin up an instance in the cloud, or pick local hardware that fits.
Advertising disclosure: we earn commissions when you shop through the links below.
Benchmark data sourced from: Hugging Face, Arena.ai, Artificial Analysis.
BB62.8 |
| 65B |
| Feb 2023 |
| 3 | text | BB56.1 | 70B | Dec 2024 |
| 4 | text | BB55.2 | 8B | Jul 2024 |
| 5 | text | CC51.0 | 8B | Apr 2024 |
| 6 | text | CC50.3 | 400B | Apr 2025 |
| 7 | text | CC49.9 | 70B | Jul 2024 |
| 8 | text | CC47.7 | 109B | Apr 2025 |
| 9 | text | CC43.7 | 70B | Apr 2024 |
| 10 | text | CC42.9 | 13B | Jul 2023 |
Oct 21
Apr 4
Apr 4
Jul 22
Jul 22
Jul 22
Apr 17
Apr 17
Jul 17
Jul 17
Jul 17
Feb 23
Composite grades across this family. Higher is better, blending benchmarks, popularity, and efficiency.
BB62.8 |
| 65B |
| Feb 2023 |
| 3 | text | BB56.1 | 70B | Dec 2024 |
| 4 | text | BB55.2 | 8B | Jul 2024 |
| 5 | text | CC51.0 | 8B | Apr 2024 |
| 6 | text | CC50.3 | 400B | Apr 2025 |
| 7 | text | CC49.9 | 70B | Jul 2024 |
| 8 | text | CC47.7 | 109B | Apr 2025 |
| 9 | text | CC43.7 | 70B | Apr 2024 |
| 10 | text | CC42.9 | 13B | Jul 2023 |
| 11 | text | CC41.8 | 405B | Jul 2024 |
| 12 | embedding | CC41.4 | 3.2B | Mar 2026 |
| 13 | text | CC40.1 | 7B | Jul 2023 |
| 14 | text | DD37.9 | 70B | Jul 2023 |
Vultr
GPU cloud with hourly and monthly plans, starting at $0.50/hr.
