Llama 3.3 70B Instruct
Wandb textwandb/meta-llama/Llama-3.3-70B-Instruct
Input
$0.7100
per 1M input tokens
Output
$0.7100
per 1M output tokens
Cache read
n/a
per 1M cached read tokens
Cache write
n/a
per 1M cache write tokens
Context window
128,000
Max output
128,000
Effective date
Aug 28, 2026
Estimate a workload
Enter token counts to see the cost at this model's rates.
Estimated cost:
$0.00
Price history
blended $ per 1M tokens (3:1)| Effective | Input | Output | Blended |
|---|---|---|---|
| Aug 28, 2026 | $0.71 | $0.71 | $0.71 |
| Jun 15, 2026 | $71,000.00 | $71,000.00 | $71,000.00 |
Llama 3.3 70B Instruct on other providers
Hyperbolic
· $0.1200 in / $0.3000 out
Nebius
· $0.1300 in / $0.4000 out
Novita
· $0.1350 in / $0.4000 out
Crusoe
· $0.2000 in / $0.2000 out
Inferenceapis
· $0.2000 in / $0.5000 out
Inferenceapis
· $0.2000 in / $0.5000 out
Nscale
· $0.2000 in / $0.2000 out
Openrouter
· $0.2200 in / $0.5000 out
Deepinfra
· $0.2300 in / $0.4000 out
Gmi
· $0.2500 in / $0.7500 out
Azure OpenAI
· $0.7100 in / $0.7100 out
Scaleway
· $0.9000 in / $0.9000 out
Other Wandb models
wandb/openai/gpt-oss-120b
· $0.1700/Mtok out
wandb/openai/gpt-oss-20b
· $0.1300/Mtok out
wandb/ibm-granite/granite-4.1-8b
· $0.1000/Mtok out
wandb/JetBrains/Mellum2-12B-A2.5B-Instruct
· $0.1000/Mtok out
wandb/OpenPipe/Qwen3-14B-Instruct
· $0.2200/Mtok out
wandb/ibm-granite/granite-4.2-8b
· $0.1500/Mtok out
Source: litellm. Confirm against the provider's official pricing before relying on these figures.