更新日志

Use jev-1.14.0 or jev-preview to evaluate text and images for classification, scoring, and object localization with Box coordinates.

PNG and JPEG images are supported, with structured results your application can use directly. OmniaKey handles the image activation header automatically. Pricing remains the same as existing Jev models.

👉 Read the integration guide and examples
https://docs.omniakey.com/en/api/jev

Cached input costs just half as much as GPT-6 Sol.

Full GPT Image 2.5 and Nano series are now supported, plus Seedream 5.0 Pro/Lite, Qwen 3.0, Wan 2.7 and more.

GPT Image from $0.068/image, Nano from $0.039/image.

Just tell your AI agent to use any image model on OmniaKey. Example: "Use OmniaKey with GPT or NanoBanana Image to make a cinematic product photo. Read the doc first:
https://docs.omniakey.com/media/image"

Hi all,
DeepSeek V4.1 Flash is now live. Request it via model name like deepseek-v4.1-flash or deepseek-flash

Model ID gpt-6-astra. $0.70 input, $3.50 output, $0.07 cached input per 1M tokens.
GPT-5.6 Sol, for comparison: $0.35 / $2.20 / $0.038.