# BytePlus ModelArk — AI Compare Ninja

> BytePlus API platform for selectable AI models, inference, fine-tuning and evaluation.

## Overview
- **Category:** Development & Code (Agent & RAG frameworks)
- **Status:** Active · checked - Active API model platform. Model version and availability depend on market and model.
- **Paid from:** 0.60 USD / 1M input tokens / usage-based (Kimi K2, online inference)
- **Free access:** not published (Public trial quota depends on model. No universal amount is confirmed.)
- **Website:** https://www.byteplus.com/
- **Canonical URL:** https://www.aicompare.ninja/en/tool/byteplus-modelark

## What it does
- Model-dependent API prices rather than one universal tariff
- Kimi K2: input 0.60 USD/M tokens, cache hit 0.12 USD/M, output 2.50 USD/M; cache storage and batch separate
- Customer data not used for own model training without prior authorization per data-processing page

## In brief
BytePlus ModelArk is an API platform with model-dependent pricing and availability. Current Kimi K2 documentation lists 0.60 USD per million input tokens and 2.50 USD per million output tokens, not a general price for all models. The data-processing page says customer data is handled only under customer instructions; the specific training promise for BytePlus video models does not automatically apply to every other model.


## Limits
According to the current data-processing documentation, ModelArk processes customer data as a processor only under customer instructions, in Malaysia, Indonesia and/or the EU/EEA; no own model training without prior authorization. Inputs and outputs triggered by content filters may be stored for 180 days in Malaysia. BytePlus video-generation models additionally promise no base-model training on input or output without separate consent. Do not extend that specific promise to every third-party or text model. Availability and model version can differ by country; the point of purchase controls.

## Pricing tiers
| Tier | Price | Period | What you get |
| --- | --- | --- | --- |
| Kimi K2, online inference | 0.60 USD / 1M input tokens | usage-based | Kimi K2: input 0.60 USD/M tokens, cache hit 0.12 USD/M, output 2.50 USD/M; cache storage and batch separate |
| Other ModelArk models | contact sales | custom | Check price, quota and availability on each model card; do not infer a universal price |

## Sources
All details regarding plans, limits, and features are based on official vendor documentation and the primary sources verified below.

- https://docs.byteplus.com/en/docs/modelark/1747867
- https://docs.byteplus.com/en/docs/modelark/1108216
- https://docs.byteplus.com/en/docs/modelark/BytePlus_ModelArk_Data_Processing
- https://docs.byteplus.com/en/docs/ModelArk/Specific_Terms_for_the_BytePlus_Video_Generation_Model_Services
- https://docs.byteplus.com/en/docs/ModelArk/availability

---
*Source: AI Compare Ninja (2026) - https://www.aicompare.ninja/en/tool/byteplus-modelark*