#minimax-h3

6 articles

MiniMax H3 Now Makes Images: Text-to-Image and Image Edit APIs Live on WaveSpeedAI

The open-weights MiniMax H3 family now includes two still-image endpoints on WaveSpeedAI: cinematic text-to-image at 1K and 2K from $0.02, and a reference-guided image editor that keeps a person's identity across new scenes, outfits, and styles from $0.03. LoRA variants are available for both.

2026-09-066 min read

MiniMax H3 Adds Video Edit, Video Extend, and a 540p Tier

The open-weights MiniMax H3 family on WaveSpeedAI now covers video editing, video extension, and a 540p resolution tier at 1.5x the 480p price. Every parameter, every price, and when to use which.

2026-09-055 min read

Sora 2 API Shuts Down September 24: Migrate to Seedance 2.5, Wan 3.0, or MiniMax H3

OpenAI removes the Videos API and every Sora 2 model on September 24, 2026 with no named replacement. Here is how to map each Sora 2 request field to Seedance 2.5, Wan 3.0, and the open-weights MiniMax H3, with real per-clip prices and a three-week plan.

2026-09-058 min read

Sora 2 to Seedance 2.5, Wan 3.0, or MiniMax H3: The Request Bodies and Python Code

Exact request bodies for moving Sora 2 text- and image-to-video calls to Seedance 2.5, Wan 3.0, and MiniMax H3, plus a size mapper and a Python submit-and-poll client.

2026-09-057 min read

MiniMax H3 AI Video Model: What Can We Expect?

What could MiniMax H3 bring to AI video generation? A practical look at motion control, reference images, consistency, and workflow fit.

2026-07-285 min read

MiniMax H3 AI Video Model: What Could Make It Impressive?

A practical look at the signals that could make MiniMax H3 stand out for creators and developers, from control and stability to production workflows.

2026-07-286 min read