PixAI

PixAI is a generative AI platform developed by Mewtant Inc. for anime, manga, and illustration workflows. It provides text-to-image generation, reference-based generation, image editing, community model and LoRA sharing, and online LoRA training. The service is accessible through the web and mobile applications.^[1][2]

Background

According to the company's account of its history, PixAI began operating in October 2022.^[3] Independent reporting by ITmedia in September 2026 documented the service's exhibition at Tokyo Game Show, where Metanomaly demonstrated a business-facing interface and printed images generated with the consumer service.^[4]

Models and image generation

PixAI offers multiple model families rather than one architecture for all tasks. Its documentation describes SDXL models alongside the proprietary Tsubaki DiT family.^[5] In September 2026, the company announced the official launch of Tsubaki.3. Its current documentation identifies Tsubaki.3 as the flagship image model.^[6][7]

The company describes Tsubaki.3 as a multimodal foundation model for anime illustration, manga, and webtoon generation. Its technical report separates text-to-image generation from three image-conditioned task families: local editing, reference generation, and reference editing. A distilled offering, Tsubaki.3 Flash, uses fewer generation steps.^[8]

The product's documented workflows include character design sheets, manga pages, text within images, and reference-based editing. A Color Palette feature accepts preset or custom color schemes.^[7] These are manufacturer descriptions of capabilities, not independent evidence of superiority over competing models.

LoRA and community features

Users can search for shared LoRAs, adjust their weights and trigger words, and train LoRAs from uploaded datasets. Compatibility depends on the base-model architecture: an SDXL LoRA is not compatible with the DiT-based Tsubaki.2 workflow.^[9]

Editing, video, and Studio

PixAI Edit and Reference Pro use uploaded images and natural-language instructions. Reference Pro focuses on combining multiple reference images, while PixAI Edit supports iterative conversational editing.^[10]

PixAI also offers image-to-video workflows. A September 2026 announcement introduced the Tsubaki Video family.^[11] PixAI Studio organizes media and generation steps in a node-based workspace. Its 4 September 2026 update introduced Sequence Editor for timeline-based media assembly and Egaki for layer separation and region-specific image editing.^[12]

Access and limitations

PixAI provides free access with credits and paid memberships. Generation costs vary with models and settings.^[13] Content is subject to the service's moderation policies; availability and restrictions can differ between web and mobile interfaces. Product capabilities, prices, and limits may change after the dates of the sources cited here.

Continue reading

• Adobe FireflyAdobe Firefly is a family of creative generative AI models developed by Adobe and integrated across its product ecosystem.Also in Image Generation
• Recraft AIRecraft AI is an AI image generation platform built for professional designers, creative teams, and brand-focused workflows.Also in Image Generation
• HHiggsfield AI InfluencerHiggsfield AI Influencer is a visual character-building tool within Higgsfield AI. It creates synthetic personas from appearance settings, with an optional photograph as a facial reference.Also in Image Generation
• BBest AI Image GeneratorsAs of July 2026, the best all-around AI image generator is GPT Image 2 from OpenAI.Also in Image Generation
• Stable DiffusionStable Diffusion is a family of generative image models that can synthesize and edit images from text and other conditions.Also in Image Generation
• DALL-EDALL-E is a family of text-to-image systems developed by OpenAI that generates images from natural-language descriptions.Also in Image Generation

More to read

• Stability AI
• Midjourney
• Diffusion Transformer (DiT)
• Adobe
• Black Forest Labs
• Imagen (text-to-image model)

Topics:AI Tools & ProductsGenerative AIImage Generation

References

1. ^Mewtant, company product page: mewtant.io
2. ^PixAI documentation, Introduction: docs.pixai.art/...introduction
3. ^PixAI, "PixAI Creator Lab (PCL) Launches," 11 May 2026: blog.pixai.art/...pixai-creator-lab-pcl
4. ^ITmedia NEWS, "東京ゲームショウと生成AI　ゲームの祭典で商談見込むベンダーたち," 19 September 2026: itmedia.co.jp/...2000001639
5. ^PixAI documentation, Other models: docs.pixai.art/...other-models
6. ^PixAI, launch announcement, 16 September 2026: pixai.art/...2056884675156132291
7. ^1 ^2PixAI documentation, Tsubaki.3: docs.pixai.art/...tsubaki3
8. ^PixAI, Tsubaki.3 Technical Report: pixai.art/...report
9. ^PixAI documentation, LoRA usage: docs.pixai.art/...lora-usage
10. ^PixAI documentation, Editing models: docs.pixai.art/...editing-models
11. ^PixAI, fourth anniversary announcement: pixai.art/...2056964974369939540
12. ^PixAI, Studio update, 4 September 2026: blog.pixai.art/...eate-long-videos-separate-layers
13. ^PixAI documentation, Membership: docs.pixai.art/...membership

Show all 13 references