97AI Product Updates & Changelog

Seedream 5 Pro Layer Decomposition is now available through the 97AI model relay.

🌟 Key features

• Automatically separates complex images into independent editable layers, including background, foreground and key visual elements.

• Preserves detail, structure and layer consistency for more precise downstream editing.

• Analyzes complex compositions and extracts visual components with high accuracy.

💰 Pricing

• 1K / 1.5K: 7.35 credits per output image.

• 2K: 14.7 credits per output image.

• Pricing is based on generated output images; each separated layer, including the base image, counts as one output image.

📚 The model detail page, API docs and pricing rows are now live.

New media models and expanded generation workflows have been added to the unified 97AI API.

🧩 Newly available

• Wan 3.0 Video — text-to-video, first-frame / first+last-frame, all-in-one references, audio references, files and links.

• Qwen Image 3.0 — text-to-image and image-to-image, with Standard and Pro variants.

• Seedance 2.5 — one flexible endpoint for no-video generation or image, video and audio references, with synchronized audio and clips up to 30 seconds.

💰 Pricing

• Wan 3.0 Video — 480P 7.5 credits/s, 720P 15 credits/s, 1080P 30 credits/s.

• Qwen Image 3.0 — from 4.8 credits/image; Pro 1K 6.4 credits/image, Pro 2K 12 credits/image.

• Seedance 2.5 480P — 17.85 credits/s with video input; 29.4 credits/s without video input.

• Seedance 2.5 720P — 39.9 credits/s with video input; 66.15 credits/s without video input.

📚 Detail pages, docs pages and pricing rows are now live for the new models.

Higher-quality and more cost-efficient video generation options have been added to the Models Market.

✨ Highlights

• Grok now supports 1080P video generation.

• Grok Upscale now supports 480P → 1080P and 720P → 1080P.

• MiniMax H3 now supports 768P video generation.

💰 Pricing

• Grok 1080P — 8 credits (≈ $0.04).

• Grok Upscale 480P → 1080P — 30 credits (≈ $0.15).

• Grok Upscale 720P → 1080P — 20 credits (≈ $0.10).

• MiniMax H3 768P video generation / video input — 22.5 credits/s (≈ $0.1125/s).

• MiniMax H3 image input — 11 credits/image (≈ $0.055/image).

🚀 High-tier top-ups still include a 10% credit bonus.

Faster sign-in and easier model discovery across the platform.

✨ Highlights

• Sign in with Google — one tap from the login window, no password required. Email + password still works.

• Model comparison pages — 135 side-by-side "A vs B" pages comparing price, provider and availability across our video, image and chat models.

• New guides: pay with USDC crypto, use AI without a credit card, GDPR-compliant access for the EU, and a China usage guide.

More generation options added to the Models Market.

🧩 Newly available

• PixVerse — text-to-video, image-to-video, reference-to-video, extend and transition.

• Seedream v3 — image generation.

• Gemini 3.1 Flash TTS — text to speech.

🚀 All callable through the same relay API with pay-per-use credits.

This release adds a dedicated detail page for every model with an inline playground and copy-paste API snippets.

✨ Highlights

• API keys can now be scoped to specific models — the relay enforces the allow-list and returns 403 otherwise.

• Billing page: real balance, transaction history, CSV export, and test-mode top-ups.

• API Usage and Logs now read live from your account activity.

The relay now exposes an OpenAI-compatible chat endpoint for every supported LLM.

🧩 Available families

• Claude — Opus, Sonnet, Haiku, Fable

• GPT-5.6 — Sol, Terra, Luna

• Gemini — 2.5/3 Pro & Flash

• Grok — 4.x

🚀 Call POST /api/chat with { model, messages } and your bearer key.

The Models Market opened with the full catalog of image, video, music, speech and chat models.

Every model is callable through one relay API, billed with pay-per-use credits — no per-provider accounts required.