GPU fit checker

Static tables per VRAM tier. Agents: /api/fit?vram=12.

8 GB — RTX 4060

✅ ESRGAN model upscaling ~0.07 GB (estimate)
✅ SD 1.5 inpainting ~4.26 GB (estimate)
✅ AnimateDiff-Evolved text-to-video ~6.08 GB (estimate)
✅ SDXL Base + Refiner text-to-image ~6.94 GB (estimate)
⚠️ Wan 2.1 text-to-video (1.3B) ~8.19 GB (official) — try: Stay at 480p
⚠️ LTX-Video text/image-to-video ~9.37 GB (estimate) — try: fp8 T5
⚠️ Flux.1 Schnell 4-step text-to-image ~11.89 GB (estimate) — try: fp8 weight_dtype
⚠️ Wan 2.1 image-to-video (14B) ~16.4 GB (estimate) — try: Use GGUF quantized weights via ComfyUI-GGUF (city96)
⚠️ Flux.1 Dev text-to-image ~23.8 GB (estimate) — try: weight_dtype fp8 halves model memory (per source)
⚠️ Flux Fill inpainting & outpainting ~23.8 GB (estimate) — try: fp8 weight_dtype
⚠️ Flux Kontext image editing ~23.8 GB (estimate) — try: fp8 scaled Kontext model
⚠️ HunyuanVideo text-to-video ~45 GB (official) — try: weight_dtype fp8

✅ ESRGAN model upscaling ~0.07 GB (estimate)
✅ SD 1.5 inpainting ~4.26 GB (estimate)
✅ AnimateDiff-Evolved text-to-video ~6.08 GB (estimate)
✅ SDXL Base + Refiner text-to-image ~6.94 GB (estimate)
✅ Wan 2.1 text-to-video (1.3B) ~8.19 GB (official)
✅ LTX-Video text/image-to-video ~9.37 GB (estimate)
✅ Flux.1 Schnell 4-step text-to-image ~11.89 GB (estimate)
⚠️ Wan 2.1 image-to-video (14B) ~16.4 GB (estimate) — try: Use GGUF quantized weights via ComfyUI-GGUF (city96)
⚠️ Flux.1 Dev text-to-image ~23.8 GB (estimate) — try: weight_dtype fp8 halves model memory (per source)
⚠️ Flux Fill inpainting & outpainting ~23.8 GB (estimate) — try: fp8 weight_dtype
⚠️ Flux Kontext image editing ~23.8 GB (estimate) — try: fp8 scaled Kontext model
⚠️ HunyuanVideo text-to-video ~45 GB (official) — try: weight_dtype fp8

✅ ESRGAN model upscaling ~0.07 GB (estimate)
✅ SD 1.5 inpainting ~4.26 GB (estimate)
✅ AnimateDiff-Evolved text-to-video ~6.08 GB (estimate)
✅ SDXL Base + Refiner text-to-image ~6.94 GB (estimate)
✅ Wan 2.1 text-to-video (1.3B) ~8.19 GB (official)
✅ LTX-Video text/image-to-video ~9.37 GB (estimate)
✅ Flux.1 Schnell 4-step text-to-image ~11.89 GB (estimate)
⚠️ Wan 2.1 image-to-video (14B) ~16.4 GB (estimate) — try: Use GGUF quantized weights via ComfyUI-GGUF (city96)
⚠️ Flux.1 Dev text-to-image ~23.8 GB (estimate) — try: weight_dtype fp8 halves model memory (per source)
⚠️ Flux Fill inpainting & outpainting ~23.8 GB (estimate) — try: fp8 weight_dtype
⚠️ Flux Kontext image editing ~23.8 GB (estimate) — try: fp8 scaled Kontext model
⚠️ HunyuanVideo text-to-video ~45 GB (official) — try: weight_dtype fp8