Model Comparisons
39 comparisons · drag the sliders to diff A/B · built 2026-10-28 11:57
H3 Regen-2K Blind
FlashVSR+ vs FlashVSR v1.1 — VideoLQ ×4
Qwen-Image 2.1 vs FLUX.3 Image Klein
Video description: Marlin vs MiniCPM-o vs AVF
Animate model comparison
SoL-Refiner Reference Check
H3 FlashGen Refined
H3 Shipped vs FlashGen
MiniMax-H3: native 1344×768 vs 896×512 draft + SoL-Refiner (H3 vs LTX-2.3) — blind
SoL-Refiner on MiniMax-H3 — LTX-2.3 one-step vs LTX-2.5 H3 refiner (blind, 1080p)
Qwen-Image 2.1 edit (int8, RTX 5090) — raw vs PE-I2I rewritten instructions
Qwen-Image 2.1 (int8, RTX 5090) — raw vs PE-rewritten prompts
MiniMax-H3 — 8-step vs 4-step turbo (HTTP A/B)
MiniMax-H3 ref2va — which encoder binds references better?
MiniMax-H3 — does a better text encoder show?
MiniMax-H3 — which turbo LoRA?
LTX-2.5 — INT8-ConvRot vs BF16
MiniMax-H3 — sampling steps vs cache stride
MiniMax-H3 — distillation LoRA vs fewer steps
MiniMax-H3 — image-to-video prompt format
MiniMax-H3 — text-to-video prompt format
Super-resolution ×4 — SeedVR2 vs Real-ESRGAN (Set14, RTX 5090)
Super-resolution ×2 — SeedVR2 vs Real-ESRGAN (Set14, RTX 5090)
Super-resolution ×4 — SeedVR2 vs Real-ESRGAN (DIV2K valid, RTX 5090)
Super-resolution ×2 — SeedVR2 vs Real-ESRGAN (DIV2K valid, RTX 5090)
Video SR ×4 — SeedVR2 vs FlashVSR (VideoLQ, RTX 5090)
Qwen-Image-Edit-2511 — FP8 variants (img2img, RTX 5090)
Lens — 5090 native vs RTX 4080 SUPER (GGUF GPT-OSS TE + FP8/Sage/fused-adaLN DiT)
FLUX.2 Klein — 4B vs 9B (bf16 vs FP8) txt2img on RTX 5090
FLUX.2 Klein — 4B vs 9B (bf16 vs FP8) img2img on RTX 5090
FLUX.2 Klein — bf16 vs FP8 vs NVFP4 on RTX 5090
Krea-2-Turbo — bf16 vs ISK-fp8 (W8A8) — RTX 5090
Krea-2-Turbo bf16 — resolution comparison (RTX 5090)
JoyAI-Image-Edit-Plus — bf16 showcase (RTX PRO 6000)
Flux.2 Klein vs Qwen-Image-Edit-2511 — img2img (RTX 5090)
Flux.2 Klein vs Qwen-Image-Edit-2511 — img2img @ 1 MP (RTX 4090 / SM89)
Flux.2 Klein vs Qwen-Image-Edit-2511 — img2img @ native 1 MP (RTX 5090)
Qwen-Edit-2511 @ 1 MP — resolution + SageAttention A/B (RTX 5090)
Boogu-Image-0.1-Turbo FP8 — resolution comparison (RTX 5090)