Meta Models & Pricing
Open-weight Llama models, typically served via hosting providers; pricing shown reflects a representative host.
| Llama 4 Maverick | $0.2200 | $0.8500 | — | 1,000,000 |
| Llama 4 Scout | $0.1100 | $0.3400 | — | 10,000,000 |
| Llama 3.3 70B | $0.6000 | $0.6000 | — | 128,000 |
| Llama-3.3-70B-Instruct | $0.1350 | $0.4000 | — | 128,000 |
| Llama-4-Maverick-17B-128E-Instruct-FP8 | $0.2000 | $0.8000 | — | 128,000 |
| Muse Spark 1.1 | $1.25 | $4.25 | $0.1500 | 1,000,000 |
| Muse Spark 1.2 | $1.25 | $4.25 | $0.1500 | 1,048,576 |
| Muse Spark 1.2 Contributor | $0.1000 | $0.2000 | $0.00200 | 1,048,576 |
| Llama 3.1 70B Instruct | $0.4000 | $0.4000 | — | 131,072 |
| Llama 3.1 8b Instruct | $0.0250 | $0.0250 | — | 16,000 |
| Llama 3.2 11B Vision Instruct | $0.0550 | $0.0550 | — | 16,000 |
| Llama 3.2 1B Instruct | $0.0100 | $0.0100 | — | 16,000 |
| Llama 3.2 3B Instruct | $0.0200 | $0.0200 | — | 16,000 |
| Llama 3 70B Instruct | $0.5100 | $0.7400 | — | 8,192 |
| Llama 4 Scout 17B 16E Instruct | $0.1000 | $0.3000 | — | 327,680 |
| Muse Glimmer 30B | $0.2000 | $0.8000 | $0.0500 | 131,072 |
Source: Meta pricing. Prices are re-checked daily and are for planning only — confirm current rates on the provider’s page. Image, audio and video models aren’t priced per token, so they show their native unit (per image, minute or 1M characters) instead of input/output token rates — open the model for the full breakdown.