Draw Things: Offline AI Art
相片與影片
免費 · App 內購買
Unleash your imagination with Draw Things, the powerful offline AI creation tool for your iPhone or iPad. Create stunning images privately, right on your device – no internet or subscriptions needed for local creation!
• Your Private AI Studio: Create art 100% offline. Your prompts and images stay securely on your device.
• Powerful AI Diffusion Engine: Harness the latest models like SDXL, FLUX.1, Wan 2.2, Qwen Image, Z Image, FLUX.2 and LTX-2.3 for incredible results.
• Pro-Level Control: Master image creation with on-device LoRA training, full ControlNet support, inpainting, outpainting, and pose editing.
• Unlimited Customization: Import your favorite community-made diffusion models and LoRAs effortlessly.
• Free Local Creation: Create limitlessly on your device without ads or token limits. (Optional Cloud Compute available).
• Optimized for Apple: Experience blazing-fast performance, especially on M1, M2, and M3 Apple Silicon Macs. Native support for iPhone, iPad, and Mac.
• Complete Creative Suite: Explore text-to-image, image-to-image, AI video creation, infinite canvas, scribble-to-image, and more.
Perfect for AI enthusiasts, artists, designers, and creators seeking unparalleled control over AI creation. New to AI image / video creation? Dive into our comprehensive Wiki and join our vibrant Discord community for guides, tips, and support!
EULA: https://www.apple.com/legal/internet-services/itunes/dev/stdeula/
Privacy Policy: https://drawthings.ai/policy/
更多 * Add interrogator model "Qwen 3.5 4B".
* Add dedicated mode selector "Auto" / "Generation" / "Edit".
* Support "Sign in with Google".
* Some other small visual updates.
1.20260518.2 5月23日
* Hotfix for new segmented downloader issue.
1.20260501.0 5月4日
* Support SeedVR2 as one of the diffusion model for 1-step upscale / restoration work.
* Fix Anima LLM adapter related issues.
* Switch to better newer causal / mask attention kernels.
* Support export LoRAs for Qwen, Z Image, FLUX.2 and ERNIE Image models.
* Imported Z Image models now runs at BF16 precision.
1.20260430.0 5月1日
* Added support for ERNIE Image, including LoRA import, training and model import.
* Added support for Anima Preview 3, including LoRA import, training and model import.
* Updated LTX-2.3 [distilled] to LITX-2.3 [distilled] 1.1.
* Fixed crash related to Z-Image LoRA training.
* For M5 / A19 Pro, now defaults to ANE+NAX hybrid inference mode.
1.20260418.1 4月20日
* Add 8-bit S support for M4 / A18 Pro devices, reaching up to 2x speed-up for these device.
* Support LoRA training for FLUX.2 [klein] 4B / 9B / Z Image / Qwen Image as base model.
* Fix audio sync issue during in-app playback.
* Fix some 8-bit S model generation issues on M5 / A19 Pro.
1.20260410.1 4月11日
* Added a new quantization variant: 8-bit S, which on A19 Pro / M5 is about 1.5x to 2x faster than Exact or 8-bit variant.
* Fixed issue when importing models, it was no longer respecting External Folder selection.
* Support Bridge Mode to Cloud Compute with Bring Your Own LoRAs.
* Some language updates to our tiers, now it is Local Only, Free, Draw Things+.
1.20260330.0 3月31日
* Support export video as HEVC;
* Fix new History item menu issue.
1.20260323.0 3月24日
* Fix FLUX.2 [klein] 9B KV cannot use batch size or text guidance != 1 correctly.
* Better Hires Fix logic specifically for LTX-2.3 so it will be more automatic to find x2 or x1.5 resolutions.
* New history view on iPad with more log-like behavior.
1.20260322.0 3月23日
* Fix FLUX.2 [klein] 9B KV cannot use batch size or text guidance != 1 correctly.
* Better Hires Fix logic specifically for LTX-2.3 so it will be more automatic to find x2 or x1.5 resolutions.
* New history view on iPad with more log-like behavior.
1.20260320.1 3月23日
* Support LTX-2.3 for crisp clean audio and video generation;
* Support FLUX.2 [klein] 9B KV for up to 2x speed-up during image editing;
* Speed up video model's decoding performance.
1.20260314.0 3月16日
* Fix metadata for LTX-2 models.
* Fix Z-Image LoRA trained by OneTrainer import issues.
* Revert to index to the last frame for generated videos to facilitate some existing scripts workflow.
* Fix Compute Units calculation issues for FLUX.1.
1.20260304.0 3月5日
* Fixed a crash when generate LTX-2 video and enabled "Save Generated Media to";
* New Compute Units formulation that encourages video generation with Cloud Compute;
* New audio playback UI for video with audio.
1.20260303.0 3月4日
* Support LTX-2 models for video generation with audio builtin, support both I2V and T2V mode, as well as importing their LoRAs;
* Fixed FLUX.2 issues related to loading LoRAs as side-path;
* Fixed FLUX.2 issues with tiled diffusion.
1.20260228.0 3月3日
* Support Z Image Base series models;
* Support high quality preview on selected devices for Z Image, FLUX.1, HiDream, Qwen Image, FLUX.2;
* Fix importing some Z Image LoRAs.
1.20260207.0 2月11日
* Support FLUX.2 [klein] series of models (4B, 9B, 4B Base, 9B Base);
* Support both import LoRAs and import models for FLUX.2 [klein] series.
1.20260120.0 1月22日
* Support Qwen Image 2512 in FP16 and inside Official Models.
* Improve realism on Qwen Image models by fixing a precision problem.
1.20260105.0 1月7日
* Added support for Qwen Image Edit 2511;
* Added support for Qwen Image Layered.
1.20251223.0 28/12/2025
* Add support for FLUX.2, including model import and import LoRAs for FLUX.2 models.
1.20251219.0 22/12/2025
* Improve compatibility with Z Image Turbo LoRAs and finetunes.
* Fix OS 26.2 specific issue with neural accelerators.
1.20251212.0 15/12/2025
* Add support for Z Image Turbo, including model importing and LoRA support.
* Turn off automatic fallback to local rendering if remote rendering failed.
1.20251207.0 09/12/2025
Hotfix a bug in 1.20251124.0 that stopping a remote generation would trigger a local generation.
1.20251201.0 02/12/2025
* Support converting to 8-bit or quantized bit for models from the app.
* More error message on when Cloud Compute failed.
* Some minor UI fixes.
1.20251124.0 01/12/2025
* Fix performance regressions on OS 26 for pre-A19 Pro era devices;
* Use Neural Accelerators for BF16 on A19 Pro / M5 devices;
* Fix cold start slowness for Neural Accelerators;
* LoRA import now respects your version selection more.
1.20251117.1 18/11/2025
* Metal FlashAttention v2.5 w/ Neural Accelerators (preview) with breakthrough performance on M5 iPad / A19 Pro iPhones;
* Support import AuraFlow models;
* More compatibility improvement on importing Qwen Image LoRAs;
* Some minor UI adjustments.
1.20251107.1 10/11/2025
* Fix SDXL training on M3 / M4;
* Support import Wan 2.2 5B models and LoRAs.
1.20251014.0 15/10/2025
* Add interrogator model "Qwen 3.5 4B".
* Add dedicated mode selector "Auto" / "Generation" / "Edit".
* Support "Sign in with Google".
* Some other small visual updates.
更多 版本1.20260518.2 5月23日
不會與你連結的資料 開發者可能會收集以下資料，但不會連結你的個人身份：
