Meta Llama 3.1 70 B Instruct pricing
Source: litellm · updated 2026-10-27
Short answer: Meta Llama 3.1 70 B Instruct costs $0.130 per million input tokens and $0.400 per million output tokens. That is a 3:1 blended rate of $0.198 per million tokens, ranking it #165 of 704 tracked models by input price.
Price breakdown
| Metric | USD |
|---|---|
| Input, per 1M tokens | $0.130 |
| Output, per 1M tokens | $0.400 |
| Output รท input multiple | 3.08× |
| 3:1 blended, per 1M tokens | $0.198 |
| Context window | 128,000 tokens |
What a typical workload costs
| Monthly volume | Input cost | Output cost | Total | 1M in / 250K out | $0.130 | $0.100 | $0.230 |
|---|---|---|---|---|---|---|---|
| 10M in / 2.5M out | $1.30 | $1.00 | $2.30 | ||||
| 100M in / 25M out | $13.00 | $10.00 | $23.00 |
Other nebius models
| Model | Input | Output |
|---|---|---|
| Qwen 2.5 Coder 7 B | $0.010 | $0.030 |
| Llama Guard 3.8 B | $0.020 | $0.060 |
| Meta Llama 3.1.8 B Instruct | $0.020 | $0.060 |
| Qwen 2 VL 7 B Instruct | $0.020 | $0.060 |
| Mistral Nemo Instruct 2407 | $0.040 | $0.120 |
Compare against
Cheaper alternatives
Next cheaper option: Qwen 3.5.9 B at $0.170 per 1M input tokens. Next step up: GPT Luna Latest at $0.100 per 1M input tokens.