Skip to content
ModelCap

Meta

8 across 1 category board · 7 without a live endpoint · 15 equivalent routes · 4 superseded · ai.meta.com

Language ModelCap Index

8
Language rankModelGravityContextInput tokensOutput tokensReleased
#8Muse Spark 1.3API only38.21M$1.25$4.2520 days ago
#35Muse Glimmer 30BOpen weights37.9131K$0.30$1.20last month
#168Llama 4 MaverickGated access40.31M$0.188$0.652last year
#172Llama 4 ScoutGated access37.21.3M$0.10$0.30last year
#174Llama 3.3 70B InstructGated access42.1131K$0.10$0.32last year
#229Llama 3.1 8B InstructGated access42.9131K$0.05$0.082 years ago
#252Llama 3.2 3B InstructGated access30.1131K$0.05$0.33last year
#258Llama 3.2 1B InstructGated access18.560K$0.027$0.201last year

No live endpoint

7
Board statusModelGravityContextInput tokensOutput tokensReleased
No live endpointLlama Guard 3 1BGated access
Language
7.8FreeFree2 years ago
No live endpointLlama Guard 3 11B VisionGated access
Language
7.8FreeFree2 years ago
No live endpointLlama 3.2 90B Vision InstructGated access
Language
7.8FreeFree2 years ago
No live endpointLlama 3.2 11B Vision InstructGated access
Language
7.8FreeFree2 years ago
No live endpointLlama Guard 3 8BGated access
Language
7.7FreeFree2 years ago
No live endpointMeta Llama 3 8B InstructGated access
Language
7.7FreeFree2 years ago
No live endpointLlama 2 13b chat hfGated access
Language
7.6FreeFree3 years ago

Reviewed equivalent routes

15
Board statusModelGravityContextInput tokensOutput tokensReleased
Represented by Llama 4 MaverickLlama 4 Maverick 17B 128EGated access
Language
8.1FreeFreelast year
Represented by Llama 4 ScoutLlama 4 Scout 17B 16EGated access
Language
8.1FreeFreelast year
Represented by Llama 4 MaverickLlama 4 Maverick 17B 128E Instruct FP8Gated access
Language
8.11M$0.27$0.85last year
Represented by Llama 3.2 90B Vision InstructLlama 3.2 90B VisionGated access
Language
7.8FreeFree2 years ago
Represented by Llama 3.2 11B Vision InstructLlama 3.2 11B VisionGated access
Language
7.8FreeFree2 years ago
Represented by Llama 3.2 3B InstructLlama 3.2 3BGated access
Language
7.8FreeFree2 years ago
Represented by Llama 3.2 1B InstructLlama 3.2 1BGated access
Language
7.8FreeFree2 years ago
Represented by Llama Guard 3 8BLlama Guard 3 8B INT8Gated access
Language
7.7FreeFree2 years ago
Represented by Llama 3.1 405BLlama 3.1 405B FP8Gated access
Language
7.7FreeFree2 years ago
Represented by Llama 3.1 405BLlama 3.1 405B Instruct FP8Gated access
Language
7.7FreeFree2 years ago
Represented by Llama 3.1 405BLlama 3.1 405B InstructGated access
Language
7.7FreeFree2 years ago
Represented by Llama 3.1 405B InstructLlama 3.1 405BGated access
Language
7.7FreeFree2 years ago
Represented by Llama 3.1 70B InstructLlama 3.1 70BGated access
Language
7.7FreeFree2 years ago
Represented by Llama 3.1 8B InstructLlama 3.1 8BGated access
Language
7.7FreeFree2 years ago
Equivalent routeLlama 2 7bGated access
Language
7.6FreeFree3 years ago

Superseded versions

4
Board statusModelGravityContextInput tokensOutput tokensReleased
Superseded by Muse Spark 1.3 ContributorMuse Spark 1.2 ContributorAPI only
Language
45.91M$0.10$0.20last month
Superseded by Muse Spark 1.3Muse Spark 1.2API only
Language
43.51M$1.25$4.25last month
Superseded by Muse Spark 1.3Muse Spark 1.1API only
Language
40.81M$1.25$4.252 months ago
Superseded by Llama 3.3 70B InstructLlama 3.1 70B InstructGated access
Language
37.0131K$0.40$0.402 years ago

Other off-board models

2
Board statusModelGravityContextInput tokensOutput tokensReleased
Off boardMuse Spark 1.3 ContributorAPI only
Language
52.51M$0.10$0.2020 days ago
Off boardLlama Guard 4 12BGated access
Language
22.1164K$0.18$0.18last year

Meta models: common questions

How many AI models has Meta released?

ModelCap tracks 21 canonical Meta models as of 22 September 2026: 21 language models (8 with a public ModelCap Index position). Superseded versions and equivalent serving routes stay listed on this page, marked as such.

What is the best Meta model?

Muse Spark 1.3 is Meta's highest-ranked model as of 22 September 2026, at ModelCap Index position #8 with a score of 86.2, followed by Muse Glimmer 30B (#35). The table lists every ranked Meta model in Index order.

What is the cheapest Meta model?

Llama 3.1 8B Instruct at $0.08 per 1M output tokens ($0.05 per 1M input) as of 22 September 2026, the lowest listed price among Meta's 14 priced language models. Prices are the lowest current provider offer on the OpenRouter catalogue.

Which Meta model has the largest context window?

Llama 4 Scout, with a published context window of 1.3M tokens. Individual providers can serve less than the published maximum; each model page lists every endpoint's own limit.

Are Meta models open-weight?

16 of 21 tracked Meta models publish downloadable weights (open weights or gated access). ModelCap labels weight availability from repository metadata, not whether the training stack is open source; the licence class is shown beside each row.

What is the newest Meta model?

Muse Spark 1.3 Contributor, first listed on 2 September 2026. New Meta models appear here as soon as they pass automatic identity checks; the launches page tracks releases across every publisher.

Explore further