Pricing
You pay for the tokens you send in each request. What the API sends back (probabilities and answers) is free. Prices apply from 1 October 2026.
Every account, every month
100 million free tokens
Enough for about 200,000 requests of about 500 tokens each. Free tokens are spent first; your wallet is charged only after they run out. They reset on the 1st of each month, Vietnam time. No card and no top-up needed to start.
After the free tokens
Price per 1M tokens
- Standard request
- 500 đ
- Repeated request (cached)
- 50 đ
- What the API returns
- 0 đ
Prepay by VietQR through PayOS, from 20,000 đ. No monthly fee.
Top upRepeated requests cost 1/10
When you send a request you already sent (same account, same JSON content, same model revision), the API may return the stored result instead of running the model again. That request costs 50 đ per 1M tokens instead of 500 đ, and while you have free tokens it uses only 1/10 of its token count.
Not every repeat finds a stored result. The response tells you which price applied: "cached": true and usage.cached_input_tokens equal to the tokens billed at the repeat price.
Tiers and request limits
Your tier follows the lifetime total of your completed top-ups. It never expires and does not drop as you spend. A first top-up of 20,000 đ moves you to Tier 1: 10 times the requests per minute.
| Tier | Unlocked at lifetime top-up | Requests/min (burst) | Requests/hour | Requests/day | Concurrent | Input tokens/request |
|---|---|---|---|---|---|---|
| Free | 0 đ | 30 (10) | 1,000 | 10,000 | 2 | 4,096 |
| Tier 1 | 20,000 đ | 300 (60) | 15,000 | 300,000 | 16 | 4,096 |
| Tier 2 | 500,000 đ | 1,200 (200) | 60,000 | 1,200,000 | 64 | 4,096 |
| Tier 3 | 5,000,000 đ | 3,000 (500) | 150,000 | 3,000,000 | 128 | 4,096 |
Limits apply per account; all its API keys share them. Every tier has the same input limit, 4,096 tokens per request, which is the model’s own hard limit. The Free tier also has a daily per-IP cap shared by every free account on that IP. Over a limit, the API answers 429 with Retry-After and names the window that was hit; that request is not charged.
Top-up packs
Pay by VietQR bank transfer through PayOS. Credit usually appears within seconds on the return page; if you close it early, the system re-checks within about 5 minutes.
| Pack | You pay | You get | Buys | First top-up reaches |
|---|---|---|---|---|
| 20,000 đ pack | 20,000 đ | 20,000 đ | 40,000,000 tokens | Tier 1 |
| 100,000 đ pack+10% | 100,000 đ | 110,000 đ | 220,000,000 tokens | Tier 1 |
| 500,000 đ pack+20% | 500,000 đ | 600,000 đ | 1,200,000,000 tokens | Tier 2 |
How charging works
- Input tokens are the state, the question text and the label/level descriptions. Each response returns its own token count and cost.
- Only successful requests are charged. 4xx errors, 429 and 503 (no worker) are free.
- A request answered from a stored result ("cached": true) costs 50 đ per 1M tokens and uses 1/10 of the free tokens.
- With no free tokens and no balance left, the API answers 402 before calling the model.
- Money is stored as integer micro-đồng, so tiny requests are charged exactly.
Private deployment for companies
Need limits above Tier 3, or need TIA to run inside your own infrastructure? That is done under a separate agreement. Tell us your monthly request volume and your requirements.
Ask about private deploymentReference prices
List prices of other System One APIs from the dataset card, converted at a reference rate of 26,000 VND/USD.
| API | USD / 1M tokens | Converted / 1M tokens |
|---|---|---|
| TIA (System One API) | – | 500 đ |
| meraGPT Decider 1 | $0.03 | 780 đ |
| TypeSafe Jev 1.13.0 | $0.042 | 1,092 đ |