Models & pricing
Lume IDE ships these models built in — no API key of your own required. Each request is billed by token against your balance.
Built-in models
| Model | Capability |
|---|---|
| GPT-5.4 | Image input, reasoning |
| GPT-5.2 | Image input, reasoning, memories |
| MiniMax-M3 | Image input, reasoning |
| MiniMax-M2.7 | Reasoning |
| Kimi-K2.5 | Reasoning |
| Gemini-3.1-Pro-Preview | Image input, reasoning, memories |
| Gemini-3-Flash-Preview | Image input, reasoning, memories |
Model pricing
Rates are per million tokens (/1M tokens). Where a model shows two levels, the rate for the whole request follows the length of its input — crossing the threshold prices the entire request at the higher level, not just the overflow.
| Model | Input | Cache Read | Cache Write | Output |
|---|---|---|---|---|
| GPT-5.4 | <=272k: $5.000 >272k: $10.000 | <=272k: $0.500 >272k: $1.000 | $0.000 | <=272k: $30.000 >272k: $45.000 |
| GPT-5.2 | $3.500 | $0.350 | $0.000 | $28.000 |
| MiniMax-M3 | <=200k: $1.200 >200k: $2.400 | <=200k: $0.240 >200k: $0.480 | $0.000 | <=200k: $4.800 >200k: $9.600 |
| MiniMax-M2.7 | $0.600 | $0.120 | $0.000 | $2.400 |
| Kimi-K2.5 | $1.200 | $0.200 | $0.000 | $6.000 |
| Gemini-3.1-Pro-Preview | <=200k: $4.000 >200k: $8.000 | <=200k: $0.400 >200k: $0.800 | $0.000 | <=200k: $24.000 >200k: $36.000 |
| Gemini-3-Flash-Preview | $1.000 | $0.100 | $0.000 | $6.000 |
How billing works
- Every chat message and agent step consumes tokens; cost = tokens × the model rate above.
- Repeated context is served from cache at the Cache Read rate — typically a fifth of input. Writing to cache is free.
- Charges are deducted from your monthly Basic usage first, then any usage package, then optional On-Demand.
- Autocomplete is free on every plan and never draws from usage.
- Amounts in your account are shown in złoty; rates here are in US dollars, converted at checkout time.
See Pricing for plans and included usage. Rates last published 06/08/2026.