Llama API pricing, every Meta model per million tokens
On 10 October 2026, Meta models on this list cost from $0.02 per million input tokens (Llama 3.1 8B Instruct) to $1.25 (Muse Spark 1.3), and from $0.04 per million output tokens (Llama 3.1 8B Instruct) to $4.25 (Muse Spark 1.3).
| Model | Input | Output | Cached input | Batch in / out | Context | Your month | Listed |
|---|---|---|---|---|---|---|---|
| Muse Spark 1.3 Contributor | $0.10 | $0.20 | $0.002 | 1.05M | $9.00 | 2026-09-02 | |
| Muse Spark 1.3 | $1.25 | $4.25 | $0.15 | 1.05M | $138.75 | 2026-09-02 | |
| Muse Spark 1.2 Contributor | $0.10 | $0.20 | $0.002 | 1.05M | $9.00 | 2026-08-21 | |
| Muse Glimmer 30B Phala, cheapest of 3 hosts | $0.30 | $1.10 | $0.04 | 131K | $34.50 | 2026-08-09 | |
| Muse Spark 1.2 | $1.25 | $4.25 | $0.15 | 1.05M | $138.75 | 2026-08-05 | |
| Muse Spark 1.1 | $1.25 | $4.25 | $0.15 | 1.05M | $138.75 | 2026-07-16 | |
| Llama Guard 4 12B served by DeepInfra | $0.18 | $0.18 | 164K | $13.50 | 2025-04-30 | ||
| Llama 4 Maverick DigitalOcean, cheapest of 4 hosts | $0.1875 | $0.6525 | $0.05 | 1.05M | $21.04 | 2025-04-05 | |
| Llama 4 Scout DeepInfra, cheapest of 3 hosts | $0.10 | $0.30 | 1.31M | $10.50 | 2025-04-05 | ||
| Llama 3.3 70B Instruct Novita, cheapest of 10 hosts | $0.135 | $0.40 | 131K | $14.10 | 2024-12-06 | ||
| Llama 3.2 1B Instruct served by Cloudflare | $0.027 | $0.201 | 60K | $4.63 | 2024-09-25 | ||
| Llama 3.2 3B Instruct Parasail, cheapest of 2 hosts | $0.05 | $0.33 | 131K | $7.95 | 2024-09-25 | ||
| Llama 3.1 70B Instruct DeepInfra, cheapest of 2 hosts | $0.40 | $0.40 | 131K | $30.00 | 2024-07-23 | ||
| Llama 3.1 8B Instruct DeepInfra, cheapest of 5 hosts | $0.02 | $0.04 | 131K | $1.80 | 2024-07-23 |
9 of these 14 models are not served by Meta itself on OpenRouter, so their price is the cheapest regular price among the hosts that run them, and the host is named under the model. Compare Meta with every other provider
Common questions
How much does the Llama API cost?
On 10 October 2026, Meta models on this list cost from $0.02 per million input tokens (Llama 3.1 8B Instruct) to $1.25 (Muse Spark 1.3), and from $0.04 per million output tokens (Llama 3.1 8B Instruct) to $4.25 (Muse Spark 1.3).
What is the cheapest Llama model?
For a typical mix of three input tokens to every output token, the cheapest is Llama 3.1 8B Instruct at $0.02 per million input tokens and $0.04 per million output tokens.
What is the newest Llama model on the list?
Muse Spark 1.3 Contributor, listed on 2 September 2026, at $0.10 per million input tokens and $0.20 per million output tokens.
Which Llama model has the largest context window?
Llama 4 Scout, with 1.31M tokens of context.
Choosing a model for a production agent
Price is one input. Quality on your own cases, latency and the cost of retries decide the rest. Tell me what you are building.