Makernice
MakerNice is an API platform that provides access to 17 AI video and image generation models from ByteDance, MiniMax, Alibaba, and Kuaishou through a single key and prepaid balance. The service quotes prices upfront, locks them for 12 minutes, and automatically releases payment holds for failed or canceled runs, ensuring users only pay for successful generations.
- Artificial Intelligence
- Developer Tools
- Software Only
Funding
Founders
Founder details are not available yet.
Product
Problem
Developers integrating AI video and image generation models face fragmented access across multiple providers, unpredictable pricing, and the risk of being charged for failed or timed-out generation runs. Managing separate API keys, billing systems, and inconsistent failure handling across vendors adds operational complexity and financial uncertainty to production workflows.
Solution
MakerNice provides a unified API platform that aggregates 17 video and image generation models from ByteDance, MiniMax, Alibaba, and Kuaishou behind a single API key and prepaid US-dollar balance. The service quotes the exact price for each run upfront via a quote endpoint, locks that price for 12 minutes, and only converts the payment hold to a charge when a generation succeeds. If a run fails, times out, is canceled while queued, or is rejected by input moderation, the hold is released in full automatically. The platform also includes automatic failover to backup channels when an upstream provider goes down, and idempotent retries that return the original generation without double billing when the same Idempotency-Key is reused.
Target Audience
Primary customers are software developers and product teams building AI video and image generation features who need reliable, cost-predictable access to multiple top-tier models without managing separate vendor integrations.
Features
- Single API key and one request shape for all 17 models, with model switching via a simple `model` field change
- Upfront price quotes via `POST /v1/generations/quote` with a 12-minute price lock, billed in micro-USD
- Hold-based billing: funds are reserved at run creation and only charged on successful completion; failed, timed-out, or canceled runs release the hold automatically
- Automatic failover with per-channel circuit breakers that route requests to backup providers without code changes
- Idempotent retries using the same Idempotency-Key return the original generation result without an additional charge
- Public pricing endpoint (`GET /v1/pricing`) and a browser-based console that runs the same models at the same prices without code
- Output files retained for approximately 30 days with expiry dates; user prompts and outputs are not used for model training