Llama 3.1 70B Pricing — $0.35/M input, $0.40/M output
NVIDIA NIM / April 2026
Input $0.35/M tokens
Output $0.40/M tokens
Context Window 128K tokens
Optimized inference on NVIDIA infrastructure
Typical use cases
General-purpose inference, chat, extraction, structured output
Estimated monthly cost at scale
Assumes 50/50 input/output token split at stated daily volume.
| Daily Volume | Monthly Tokens | Estimated Monthly Cost |
|---|---|---|
| 1M tokens/day | 30M tokens | $11.25 |
| 5M tokens/day | 150M tokens | $56.25 |
| 10M tokens/day | 300M tokens | $112.50 |
vs. other NVIDIA NIM models
| Model | Input ($/M) | Output ($/M) | Context |
|---|---|---|---|
| Llama 3.1 8B | $0.05 | $0.05 | 128K |
| Mistral 7B Instruct | $0.04 | $0.04 | 32K |
Not sure if Llama 3.1 70B is the right fit for your workload? Clawback tests cheaper alternatives against your actual prompts and tells you exactly where you're overpaying.
Test if a cheaper model matches Llama 3.1 70B quality →