Pricing: Pay only for what you use
Per token, with no subscription and no per-seat fees.
Open the dashboard
$1 free for a new account
Models
USD per million tokens
| Model | Input | Cached input | Output | Cache hit rate | |
|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017DeepSeek | USD 0.15 | USD 0.01 | USD 0.60 | 99.0% | |
| GLM 5.3Z.ai | USD 1.30 | USD 0.13 | USD 4.20 | 96.3% | |
| GLM 5.3 FlashZ.ai | USD 0.11 | USD 0.03 | USD 0.45 | 98.8% | |
| Hy4 flashTencent | USD 0.06 | USD 0.01 | USD 0.25 | 99.3% | |
| Nemotron 3.5 Lightning 30BNVIDIA | USD 0.05 | USD 0.01 | USD 0.15 | ||
| Qwen3.8 27BQwen | USD 0.10 | USD 0.01 | USD 0.40 | 89.8% | |
| Ornith 1.5 35BOrnith | USD 0.10 | USD 0.01 | USD 0.40 | 98.3% | |
| DeepSeek V4.1 FlashDeepSeek | USD 0.14 | USD 0.03 | USD 0.58 | 98.7% | |
| RightNow decision modelRightNow | USD 0.19 | USD 0.019 | USD 0.00 | Plus USD 0.0001 per answered question and USD 0.00001 per answered call Full price list |
Capacity
Four ways to run
| Capacity | What it is | Price |
|---|---|---|
| Pay as you go | What it isBilled per token with no subscription, on shared capacity. | The rates above |
| Reserved capacity | What it isCapacity held for your traffic. | USD 0.25 per hour for each reserved concurrent request on Ornith 1.5 35B. Tokens at the rates above. |
| Dedicated deployments | What it isA deployment that serves your company only. | USD 9.58 per GPU hour |
| Batch | What it isA big job on idle capacity, for less. | Half of list price |
Teams
What a team gets
- No per-seat fees
- Your price does not change with the size of your team.
- Limits that grow with you
- 10,000 requests a minute per key to start, 50,000 after your first purchase, 200,000 once your purchases reach USD 200.
- A spending limit on any key
- Set a monthly budget for each key.
- Top-ups up to USD 1,000,000
- From USD 5, with optional auto-recharge. Top-ups stay valid for one year.
- Zero data retention by default
- Request and response content is never used for training. Data retention
- Prompt caching built in
- Caching is automatic and best effort, with no header or parameter required.