Per-token pricing for confidential inference
NEAR AI Cloud charges per token, with no minimums and no subscriptions. Input and output tokens are priced separately per 1,000,000 tokens, and the rate depends on which model you call. Every model below runs inside a hardware-secured enclave, and the attestation on each response is included in the price.
Rates are read from the public NEAR AI Cloud catalog API. Last checked .
How pricing works
- You pay per token. Input and output are priced separately, quoted per 1,000,000 tokens.
- The rate depends on which model you call. There is no platform fee on top of the model rate.
- No minimums, no subscriptions, and no seat licences on pay-as-you-go.
- Confidential execution and the hardware-signed attestation on every response are included in the rate, not billed as an add-on.
- Enterprise agreements add reserved GPU capacity, private models, volume rates and SLAs.
Price per model
| Model | Model ID for the API | Context window | Input, USD / 1M tokens | Output, USD / 1M tokens |
|---|---|---|---|---|
| DeepSeek V4 Flash | deepseek-ai/DeepSeek-V4-Flash | 1048K | $0.17 | $0.35 |
| deepseek-v3.2 | deepseek/deepseek-v3.2 | 128K | $1.1 | $1.1 |
| GLM 5.3 Flash | z-ai/glm-5.3-flash | 1048K | $0.15 | $0.5 |
| Kimi K2.6 | moonshotai/kimi-k2.6 | 262K | $0.81 | $3.85 |
| Kimi K3 | moonshotai/kimi-k3 | 1048K | $3.3 | $16.5 |
| Qwen 3.6 35B A3B FP8 | Qwen/Qwen3.6-35B-A3B-FP8 | 262K | $0.17 | $1.1 |
| Qwen 3.8 27B | Qwen/Qwen3.8-27B | 262K | $0.44 | $3.3 |
| qwen3-32b | qwen/qwen3-32b | 128K | $0.11 | $0.46 |
| Qwen3-VL-30B-A3B-Instruct | Qwen/Qwen3-VL-30B-A3B-Instruct | 16K | $0.15 | $0.55 |
| qwen3.5-397b-a17b | qwen/qwen3.5-397b-a17b | 128K | $0.5 | $3.3 |
The table above covers confidential text models. The catalog lists the full range, each model on its own page.
Teams
Give each person their own API key and set a spend cap on it as you create the key. Usage is reported per key, so you can see what each one costs. There is no per-seat charge, so a developer who calls the API twice a month costs you two calls rather than a licence.
Enterprise
Enterprise pricing is quoted rather than listed. It covers reserved GPU capacity, custom and private models, volume rates with SLAs, on-premise or VPC deployment, and solutions engineering. Two capabilities are requested directly: dedicated deployment for reserved model capacity, and private TEE fine-tuning for training on data that never leaves an enclave.
Pricing questions
How does NEAR AI Cloud pricing work?
- You pay per token, with input and output tokens priced separately and quoted per 1,000,000 tokens. The rate depends on which model you call. There are no minimums, subscriptions or seat licences on pay-as-you-go, so a request costs the model's input rate for the prompt plus its output rate for the completion.
Does confidential compute or attestation cost extra?
- No. Every model listed on this page runs inside a hardware-secured trusted execution environment, and each response carries a hardware-signed attestation. Both are included in the per-token rate rather than billed as an add-on.
Is there a minimum spend or subscription?
- No. Pay-as-you-go has no minimums and no subscriptions. You are billed for the tokens you actually use.
Can I control what each person on my team spends?
- Yes. Create an API key for each person and set a spend cap on the key as you create it. Usage is reported per key, so you can see what each one costs. There are no per-seat charges on pay-as-you-go, so you pay for the tokens each key actually uses rather than for the number of people who have access.
How is enterprise pricing different?
- Enterprise agreements are quoted rather than listed. They cover reserved GPU capacity, custom and private models, volume pricing with SLAs, on-premise or VPC deployment, and solutions engineering.
Where do the prices on this page come from?
- They are read directly from the public NEAR AI Cloud catalog API at request time. This page is generated on the server, so the rates shown here are the current published rates rather than a hand-maintained copy.
How do I start paying for NEAR AI Cloud?
- Create an API key in the NEAR AI Cloud console at cloud.near.ai. The API is OpenAI-compatible, so you can point an existing client at the NEAR AI base URL without changing your code.