deepseek-v4.1-flash:cached- Tool calling
- JSON output
- Reasoning
- Logprobs
- Vision · up to 8 images
low cached price
- Input
US$0.30
≈ Rp5,583
- Cached input
US$0.006
≈ Rp112
- Output
US$1.20
≈ Rp22,332
- Context
- 1,048,576
Models and pricing
Pay for the tokens you use, in US dollars per 1M tokens. Top up a prepaid balance in rupiah, and each request is billed at the rate of the lane that served it.
No plans, seats, or token bundles. The same prices apply to companies and to people working alone.
Rates per 1M tokens
Rupiah amounts are estimates at the top-up rate of 4 Nov 2026, Rp18,610 per US$1. A top-up locks its rate when you create the order.
No lanes match these filters.
| Lane | Capabilities | Input | Cached input | Output | Context |
|---|---|---|---|---|---|
| DeepSeek V4.1 Flash2 lanes | |||||
deepseek-v4.1-flash:cachedlow cached price |
| US$0.30 ≈ Rp5,583 | US$0.006 ≈ Rp112 | US$1.20 ≈ Rp22,332 | 1,048,576 |
deepseek-v4.1-flash:standard7-day cacheshared cache |
| US$0.27 ≈ Rp5,025 | US$0.045 ≈ Rp837 | US$1.08 ≈ Rp20,099 | 1,048,576 |
| Qwen3.8 27B1 lane | |||||
qwen3.8-27b:standard |
| US$0.025 ≈ Rp465 | US$0.0125 ≈ Rp233 | US$1.60 ≈ Rp29,776 | 262,144 |
DeepSeek V4.1 Flash 2 lanes
deepseek-v4.1-flash:standard- Tool calling
- JSON output
- Reasoning
- Vision · up to 8 images
- Stop sequences · stop text appears in output
- Choices · n=1 only
7-day cacheshared cache
- Input
US$0.27
≈ Rp5,025
- Cached input
US$0.045
≈ Rp837
- Output
US$1.08
≈ Rp20,099
- Context
- 1,048,576
Qwen3.8 27B 1 lane
qwen3.8-27b:standard- Tool calling
- JSON output
- Reasoning
- Logprobs
- Input
US$0.025
≈ Rp465
- Cached input
US$0.0125
≈ Rp233
- Output
US$1.60
≈ Rp29,776
- Context
- 262,144
Cached input: when a request starts with the same text as an earlier one, that part is billed at the cached rate. How much of a prompt is cached depends on the lane and on your prompts.
Estimate your monthly cost
Describe one workload. Every lane is priced at once, with the arithmetic shown.
requests
tokens
tokens
million
million
%
Cached input is the start of a prompt the lane has seen before, such as a system prompt or earlier turns. It is billed at the cached rate. How much gets cached depends on the lane and on your prompts.
A month is 30 days. After a few real requests, Usage in the console shows your actual token counts.
Use this when you already know your monthly volume, for example from another provider's bill.
Fill in every field to see the cost.
Lowest for this workload
qwen3.8-27b:standard
US$39.94per month
≈ Rp743,235
US$0.000444 per request · US$0.444 per 1,000 requests
- Input 90M × US$0.025
- US$2.25
- Cached input 135M × US$0.0125
- US$1.69
- Output 22.5M × US$1.60
- US$36.00
- Total per month
- US$39.94
No lane has a context large enough for one of these requests. Check the context column in the rates above.
Every lane
Pick a lane to see its arithmetic.
The lowest price is not always the right lane. Check capabilities and context in the rates above.
This is arithmetic on the rates above, not a quote. You pay for the tokens each request really uses. The workloads are starting points to edit, not measurements. Rupiah amounts use the same top-up rate as the rates table, and a top-up locks its rate when you create the order.
How you pay
Prepaid balance
Top up from the console. Your balance is kept in US dollars and does not expire. When it runs out, requests are refused with 402 until you top up again.
Paying from Indonesia
QRIS from Indonesian bank and e-wallet apps, up to Rp10,000,000 per top-up. Every paid top-up has a PDF receipt. Tax invoices are not available yet.
Volume and contracts
Larger companies can agree a monthly volume, a deposit, or monthly billing under a contract. Paying from outside Indonesia? QRIS only works with Indonesian apps, so tell us how you would like to pay.
Talk to the Worg team