PAYG · Pay As You GoFlexible
Top up, pay per token
Your TOKS balance is only deducted for tokens you actually use. Works across all models, with no time limit.
- ✓Start from Rp1,000 — no minimum
- ✓Balance never expires
- ✓One API key for all models
Affordable AI model API with transparent rates. Pay only for tokens used — no subscription, no monthly commitment. One API key for DeepSeek, GLM, Qwen, Kimi & more.
Top up in IDR or USD — credits tracked in TOKS and converted automatically. Rates are fixed and shown before every request.
Two ways to pay: PAYG — top up per token or PACKAGE — 24-hour quota
No credit card · Cancel anytime · 14-day refund guarantee
By making a purchase or topping up balance, you agree to our Terms of Service.
Both use the same API and the same models — the only difference is how you pay.
Your TOKS balance is only deducted for tokens you actually use. Works across all models, with no time limit.
Fixed price for large quotas. Every purchase generates a dedicated API key active for 24 hours.
Every purchase generates a dedicated API key. Quotas count input and output tokens, stay active for 24 hours, and never touch your PAYG balance.
When the quota runs out or the active period ends, the API key stops working and does not automatically use your PAYG balance.
Your TOKS balance is deducted based on the tokens you use — priced per 1 million tokens, billed in rupiah. Convert to TOKS or USD instantly using the rate above.
| Model | Provider | Context | Support | Input / 1M | Output / 1M | Status |
|---|---|---|---|---|---|---|
GLM-5.2 glm-5.2 | GLM | 1M | TextStream | Rp2,000 2 TOKS · US$0.11 | Rp4,400 4.4 TOKS · US$0.24 | Active |
DeepSeek V4 Flash 0731 deepseek-v4-flash-0731 | DeepSeek | 1M | TextStream | Rp1,200 1.2 TOKS · US$0.07 | Rp2,400 2.4 TOKS · US$0.13 | Active |
DeepSeek V4.1 Flash deepseek-v4.1-flash | DeepSeek | 1M | TextStream | Rp1,200 1.2 TOKS · US$0.07 | Rp2,400 2.4 TOKS · US$0.13 | Active |
Kimi K3 kimi-k3 | Moonshot AI | 256K | TextImagesStream | Rp4,000 4 TOKS · US$0.22 | Rp5,000 5 TOKS · US$0.28 | Active |
DeepSeek V4 Pro 0813 deepseek-v4-pro-0813 | DeepSeek | 1M | TextStream | Rp2,000 2 TOKS · US$0.11 | Rp4,000 4 TOKS · US$0.22 | Active |
Qwen 8 Max qwen3.8-max | Alibaba | 200K | TextImagesStream | Rp3,000 3 TOKS · US$0.17 | Rp6,000 6 TOKS · US$0.33 | Active |
GLM 5.3 Flash glm-5.3-flash | GLM | 1000K | TextStream | Rp1,400 1.4 TOKS · US$0.08 | Rp3,000 3 TOKS · US$0.17 | Active |
GLM 5.3 glm-5.3 | GLM | 1M | TextStream | Rp2,500 2.5 TOKS · US$0.14 | Rp3,500 3.5 TOKS · US$0.19 | Active |
Per-model details: open the AI model catalog · affordable model API guide
No recurring bills or monthly commitments. Ever.
Add credit only when you need it — you're in control.
Clear rates upfront. No hidden fees, no surprises.
Encrypted API keys, per-key rate limits, SOC2-style controls.
| Capability | 9inference | Direct to providers |
|---|---|---|
| One API key for all models | ||
| Automatic failover routing | ||
| Pay per token, no subscription | ||
| OpenAI-compatible SDK | ||
| Unified IDR & USD billing | ||
| Per-key rate limits & rotation | ||
| Manage N provider accounts | — | |
| N separate invoices / month | — |
We replaced three provider SDKs with a single 9inference key. Onboarding took 10 minutes and our infrastructure bill went down.
Pay-per-token with no subscription is perfect for our research lab. We benchmark 6 models from a single endpoint.
Failover routing saved our launch. When one provider throttles, traffic moves over automatically — zero downtime.
9inference is an affordable, pay-per-token AI model API gateway. You access DeepSeek, GLM, Qwen, Kimi and other models through a single OpenAI-compatible API key, with token pricing in rupiah and no mandatory subscription.
TOKS is our internal credit unit. 1 TOKS = Rp1,000 = US$0.0553. Each model has a rate per 1 million tokens in rupiah; you only pay for tokens actually used. No subscription, no minimum.
PAYG: you top up your TOKS balance and it's deducted per token actually used — the balance never expires and works across all models. Packages: you pay once for a quota of millions of tokens with a dedicated API key active for 24 hours — ideal for heavy workloads at a fixed price. Package quota does not use your PAYG balance, and vice versa.
No. Sign up free, create an API key, and top up anytime. You can start with any amount — there's no minimum deposit.
Requests return a top-up reminder instead of failing silently. Add credit in your wallet and you're back online within seconds. Your API key and settings stay intact.
None. You only pay for tokens used at the published rates. No monthly fees, no platform fees, no per-seat fees. The price you see is the price you pay.
Yes. 9inference is OpenAI-compatible — change the model parameter in your request and your SDK stays the same. Switch between GLM, DeepSeek, Qwen, Kimi and more with a single line.
API keys are hashed at rest, credentials are encrypted, and per-key rate limits prevent abuse. We never train models on your data and never share your prompts with third parties.
Unused credit never expires. If within 14 days of your first top-up you feel 9inference isn't a fit, contact support for a full refund of your remaining balance.
Create an API key now. One endpoint, every frontier model, billed per token — starting at 1 TOKS = Rp1,000 = US$0.0553.