# DeepSeek vs Llama: API prices compared, dated and sourced

Canonical page: https://bertrande.com/en/vs/deepseek-vs-llama/
License: CC-BY-4.0, reuse with attribution and a link.

No value below was produced, estimated or completed by a language model. Every figure was read from a source we name, on the date we give.

The cheapest DeepSeek model we track, deepseek-flash, costs $1.20 per million output tokens and $0.300 per million input tokens. The cheapest Llama model, Llama 3.1 8B Instruct, costs $0.080 per million output tokens and $0.050 per million input tokens, as sold by OpenRouter. At the entry level, Llama is 15.0 times cheaper than DeepSeek on output tokens. Prices read on 11 September 2026 and 18 September 2026.

## DeepSeek models

| Model | Input $/1M | Output $/1M | Read on |
|---|---|---|---|
| deepseek-flash | $0.300 | $1.20 | 2026-09-11 |
| deepseek-v4-pro | $1.32 | $3.96 | 2026-09-11 |

## Llama models

| Model | Input $/1M | Output $/1M | Read on |
|---|---|---|---|
| Llama 3.1 8B Instruct | $0.050 | $0.080 | 2026-09-18 |
| Llama Guard 4 12B | $0.180 | $0.180 | 2026-09-18 |
| Llama 3.2 1B Instruct | $0.027 | $0.201 | 2026-09-18 |
| Llama 4 Scout | $0.100 | $0.300 | 2026-09-18 |
| Llama 3.3 70B Instruct | $0.100 | $0.320 | 2026-09-18 |
| Llama 3.2 3B Instruct | $0.050 | $0.330 | 2026-09-18 |
| Llama 3.1 70B Instruct | $0.400 | $0.400 | 2026-09-18 |
| Llama 4 Maverick | $0.1875 | $0.6525 | 2026-09-18 |


Method: https://bertrande.com/en/methodology/
Report an error: https://bertrande.com/en/corrections/