Can I generateimages, video, voices, musicon this machine?

3-stage cascade (Würstchen architecture). Strong prompt adherence, lighter than its 5.9B suggests.

Distilled few-step model. FP16 fits comfortably in 16GB. Apache-licensed.

Full (non-distilled) Z-Image. More steps than Z-Image Turbo but higher fidelity. Apache-licensed.

The workhorse. Sharp 1024px output, huge fine-tune ecosystem (Pony, Illustrious, Juggernaut).

1-step distilled SDXL. Generates in under a second on midrange GPUs. Lower fidelity than full SDXL.

Unified text-to-image and image-editing model. One model handles generate, edit, compose.

5B T2V/I2V from Tsinghua. Mid-range hardware target. 6-second clips at 720p.

The smaller Apache-licensed FLUX.2. 4B params at 1024px — fits comfortably on midrange GPUs and high-end phones.

4 B distilled from FLUX.2 32 B. Real-time generation on consumer GPUs at FLUX-grade fidelity.

Best open full-song generator with vocals. Generates 4-minute tracks in ~20 s on a 4090.

Meta's flagship MusicGen. Strong instrumental tracks. Non-commercial only.

Stereo variant of MusicGen Large. True L/R channels for richer mix.

MMDiT architecture with triple text encoders. Better text rendering than SDXL.

Improved 3.5 generation with better composition and aesthetics than SD3 Medium.

Compact next-gen DiT with strong photographic realism. Apache-licensed for commercial use.

Realtime-capable video model. Generates faster than playback on a 4090. The speed champion.

Strong photographic realism, multilingual prompts (Chinese + English). Mid-weight footprint.

Larger Parler with broader voice and prosody range. Slower, higher fidelity.

LLM-style TTS with rich emotional control. Streaming output via vLLM.

Bilingual (Chinese + English) DiT. Strong at Chinese text rendering, light to run.

Image-to-video, ~14–25 frames at 576×1024. The original consumer-friendly local video model.

Tiny T2V. The most compatible video model — fits on 8GB GPUs. Apache-licensed.

Lightweight successor with infinite-length generation. Fits on 8 GB GPUs.

Linear DiT that generates up to 4K images on as little as 8 GB. Tiny weights, massive output.

Generates speech, laughter, sighs, music — all from a text prompt. Quirky but expressive.

Prompt-controlled TTS — describe the voice in natural language ('young woman, warm, slow').

Conversational speech model — context-aware prosody for dialogue agents.

Multilingual zero-shot voice cloning. Strong on Chinese, Japanese, Korean.

Mid-tier MusicGen. Faster than Large, slightly less rich.

Sound effects, ambient textures, short instrumental loops up to 47 s.

Refined sound design model. Cleaner ambient textures, tighter SFX.

Latent-diffusion full-song generator. End-to-end. 8 GB minimum with chunked inference.

The classic. Tiny, fast, runs on almost anything. Massive ecosystem of LoRAs and fine-tunes.

Successor to SD 1.5 with native 768px training. Smaller community than 1.5 but still light on hardware.

Featherweight 0.6B DiT with T5-XXL encoder. Beautiful 4K output for the param count.

350 M model with a 1-step distilled decoder. Sub-200 ms latency for production agents.

Flow-matching TTS with strong voice cloning from a 6-second reference clip.

Multi-language voice cloning from 6 seconds of reference audio. 17 languages.

Tiny LLM-driven zero-shot voice cloning. 4 GB VRAM, runs comfortably on most GPUs.

400 M streaming TTS on a Liquid LFM2 backbone with NVIDIA NanoCodec. 3 GB VRAM end-to-end.

Tiny MusicGen for fast prototyping. Runs on a 4 GB GPU.

Live, prompt-steerable instrumental generation. Optimized for streaming.

Stability's flagship 8B MMDiT. Sharp text, strong composition. Triple text encoder (CLIP-L, CLIP-G, T5-XXL).

Truly open-source flow-matching model. The largest fully Apache-licensed image model.

4-step distilled FLUX. The fastest way to get FLUX-tier quality. Apache-licensed.

The model to beat. State-of-the-art prompt adherence and photorealism. Hungry but worth it.

Image editing variant of FLUX. Reference image + prompt → edited result.

Hybrid DiT + MoE. Beats Flux on several benchmarks. MIT-licensed for any use.

20B MMDiT from Alibaba. Best-in-class text rendering — handles paragraphs of text in images.

Edit variant of Qwen-Image. Replace, add, restyle objects via natural prompts.

Tencent's 17B image flagship. Strong at composition and Chinese text.

Massive 84B MoE image model with 13B active. Quality rivals closed-source flagships. Needs serious hardware.

Successor to FLUX.1 dev. Higher fidelity, longer context. Heavy hardware required.

Baidu's 10B image model. Strong at Chinese prompts and stylized output.

Editing variant of HiDream-I1. Same MoE backbone tuned for image editing.

Full-size Wan T2V. Strong motion, 720p output. Quantization makes it viable down to 12GB.

Image-to-video flagship. Best open I2V motion in early 2025.

Unified text + image to video. Best-in-class T2V/I2V at 5B. Fits on a 24GB GPU at FP16.

MoE flagship — 27B total / 14B active. Top open-video quality.

Successor to LTX-Video. 4K-capable on high-end hardware. Distilled variants run on 12GB.

13B T2V — closest open rival to Sora at release. Slow but cinematic.

Lighter, sharper successor. 8.3B params, 14GB minimum, runs on consumer GPUs.

10B asymmetric DiT. Ambitious motion. Apache-licensed for any use.

Pyramidal flow-matching for efficient long videos. Strong T2V quality at 2B.

Community 8.9B distillation of FLUX.1-schnell, fully Apache 2.0. Strong photographic quality with no commercial restrictions.

Apache-licensed 9B distillation of FLUX.1. Smaller and lighter than schnell while keeping FLUX-tier prompt adherence.

The larger Apache-licensed FLUX.2. 9B params, sharper detail than the 4B variant.

Human-centric video foundation model based on HunyuanVideo. Up to 12 s clips at 24 fps, 544×960. Strong at faces and motion.

Flagship infinite-length video model derived from Wan 2.1 14B. Top-tier open-source motion at 720p.

Physics-aware text-to-image — coherent geometry and lighting, tuned for sim/robotics.

World-model T2V/I2V. Generates physics-consistent motion. Heavy VRAM appetite.

Featherweight TTS that punches way above its weight. Sub-300 ms inference, runs on CPU.

Suno-style full-song generation with synchronized lyrics + vocals. Up to 5-minute tracks.