- 🖼️ High-fidelity inpainting and image editing model
- ✍️ Strong text rendering inside edited regions
- 🏢 Built on Alibaba's Qwen Image lineage
- 🎯 Editing variant of the Qwen Image 2 generator
- 🎨 Targeted region edits with prompt control
Alibaba Group is a Chinese multinational technology company founded in 1999 and headquartered in Hangzhou, Zhejiang. Originally built around e-commerce and cloud computing, Alibaba has become one of the most prolific contributors to open-weight AI research, developing the Qwen…
Explore 39 more models by Alibaba Group →Qwen Image 2 (editing) is the inpainting counterpart to Alibaba's Qwen Image 2 text-to-image model, released March 2026. Where the base model generates images from prompts, this variant specializes in altering existing ones — masking a region and regenerating it from a text instruction while preserving the surrounding composition. Its standout trait is high-fidelity text rendering, meaning words and typography placed into an edit hold up cleanly rather than dissolving into garbled glyphs, a common failure point for image editors.
Within Alibaba's image lineup it sits alongside a higher-tier Qwen Image 2 Pro editing variant for users who need more headroom, and it succeeds the earlier Qwen Edit 2511 inpainting release. For looser content boundaries, the later Qwen Edit Uncensored covers a different niche. The Qwen family on the platform spans text, vision, embeddings, and TTS, with this model anchoring the image-editing slot.
It is best suited for precise, localized edits — swapping objects, retouching scenes, adding or correcting on-image text, and compositing changes — where maintaining the original image's integrity and rendering legible text matter most.
This About section is AI-generated from public sources via VeniceStats + Venice inference, with no human editing. It may contain inaccuracies.
| Seller | Reputation↓ | Routing | $ / img | Categories | API |
|---|---|---|---|---|---|
| ▲ Apex Ant 0x73b4…e736 | 91.02 | #1 | $0.02 | image,media | openai-images |
"Best price" and the seller table are live AntSeed catalog data (advertised $/1M tokens — or $ per generated image for unit-billed image models — not settled amounts). Reputation = on-chain trust (0-100). "Routing" = the SDK's default buyer routing (what the VPR desktop app ships with): a trust ≥ 60 gate on the effective reputation, then cheapest-first among routable sellers; live failover state (per-peer cooldowns) is buyer-side runtime and not included. Model knowledge (TLDR, provider, About) via the VeniceStats enrichment layer. Advertised catalog, not the model used in any specific purchase. "Usage on AntSeed" counts only settlements whose buyers share the per-model split on-chain (metadata v2/v3, opt-in), so every usage figure is a lower bound.