Mistral Small: price on every cloud
Open weights · 24 billion parameters · 32,768-token context · licence apache-2.0
Per token, by cloud
| Cloud | Input, per 1M tokens | Output, per 1M tokens | Cheapest region |
|---|---|---|---|
| Core42 | $0.12 | $0.36 | uae |
| AWS | $1.00 | $3.00 | us-east-1 |
| Azure | — | — | not sold per token |
| OCI | — | — | not sold per token |
| Alibaba Cloud | — | — | not sold per token |
| Google Cloud | — | — | not sold per token |
Run it yourself
It needs 58.9 GB of GPU memory at bf16, the precision it ships in.
| Cloud | Cheapest instance that fits | GPUs | An hour | A month, around the clock |
|---|---|---|---|---|
| Azure | Standard_NC24ads_A100_v4 | 1 × A100 80 GB | $3.67 | $2,681 |
| Google Cloud | g4-standard-48 | 1 × RTX PRO 6000 96 GB | $4.50 | $3,285 |
| Alibaba Cloud | ecs.gn9gc.4xlarge | 1 × L20N 72 GB | $6.71 | $4,898 |
| AWS | p5.4xlarge | 1 × H100 80 GB | $6.88 | $5,022 |
At the cheapest per-token price, running it on Azure Standard_NC24ads_A100_v4 pays off above roughly 14.9 billion tokens a month.
More from Mistral
- Mistral 7B
- Mistral Large
- Mistral-7B-Instruct-v0.2
- Mistral-7B-v0.1
- Mistral-Nemo-Instruct-2407
- Mixtral-8x7B-Instruct-v0.1
Similar size, open weights
Inside the UAE: sold in-country by Core42.
List prices collected 5 October 2026, refreshed every three hours. No negotiated rates or taxes.