Catalog updated 2026-11-04

Local AI video generator for Mac

Open-weight video models, complete downloads, fit verdicts, published generation times and compatible apps, organized by Mac unified memory and NVIDIA VRAM.

video models
5
Memory tiers
13
Published measurements
33
Quick answer

Can you use a local AI video generator on a Mac?

Start with your Mac unified memory or NVIDIA VRAM, then compare the complete download, app and license. Published means a dated run by a named source; unknown means no evidence yet.

License check: review LTX-2.3, MiniMax H3 before commercial use. Local files do not remove license or territory restrictions.

Table 1

Which video models fit your Mac or NVIDIA GPU?

5 published models

Open a memory tier for its complete model table. Run counts fit and tight verdicts; unknown means no verdict is established. Runs · Stated means the linked model card, app maker or NVIDIA says that much memory runs the model; Runs · Published names the Mac or card of a dated run. On NVIDIA, Fits with offload means the main model weights fit in VRAM with a 15% margin while the text encoder and VAE run from system RAM, which is slower; Exceeds VRAM means even the main weights are too large for the card. Published Mac measurements are never used for NVIDIA verdicts or times.

Mac RAM
8 GB Mac unified memory1 unknown · 4 no
Local models for this machine
ModelFitsQualityApp
Mochi 1 previewApache License 2.0unknownComplete: calcuis GGUF Q3_K_M with Q4_0 T5 and FP8 VAE7.7 GB download#48 of 48 in text to video5,893 votessnapshot 2026-09-22Diffusers, ComfyUI
LTX-2.3LTX-2 Community License AgreementDoes not fit est.Complete: Unsloth GGUF dev Q4_K_M with GGUF Gemma 3 (ComfyUI)26.7 GB downloadNo independent score yetLightricks/LTX-2 ltx-pipelines, appautomaton/ltx-video-mlx, ComfyUI, dgrauet/ltx-2-mlx
MiniMax H3MiniMax H3 Community License AgreementDoes not fit est.Complete: Unsloth GGUF Q4_K denoiser with the Q4_K_M Qwen3-VL encoder35.5 GB download#8 of 48 in text to video11,417 votessnapshot 2026-09-22Argus-AiTeam/minimax-h3-mac (MLX streaming), antirez/h3.c (Metal), ddalcu/mlx-serve, stable-diffusion.cpp, ComfyUI
Wan2.2 A14BApache License 2.0Does not fit est.Complete: QuantStack GGUF Q4_K_M (high and low noise)26.3 GB download#42 of 48 in text to video10,458 votessnapshot 2026-09-22PocketAI Wan2.2 MLX (pocketai-wan-video), ComfyUI
Wan2.2 TI2V 5BApache License 2.0Does not fit est.Complete: QuantStack GGUF Q8_013.5 GB downloadNo independent score yetBlaizzy/mlx-video, FastVideo MLX (FastMetal), ComfyUI

Quality: blind human votes from Arena (arena.ai), CC BY 4.0, snapshot 2026-09-22. Scores are for the Arena checkpoint, not your local quantized build.

16 GB Mac unified memory2 run · 2 unknown · 2 no

Best video models for a 16 GB Mac, with the reason for each verdict

Local models for this machine
ModelFitsQualityApp
LTX-2.3LTX-2 Community License AgreementRuns · Stated16 GB unified memory · ltx-2-mlx int4 pack · ltx-2-mlx READMENo independent score yetLightricks/LTX-2 ltx-pipelines, appautomaton/ltx-video-mlx, ComfyUI, dgrauet/ltx-2-mlx
Wan2.1Apache License 2.0Runs · Stated16 GB unified memory · mlx-video 4-bit conversion · mlx-video Wan READMENo independent score yetml-explore/mlx-examples Wan2.1, ComfyUI, FastVideo MLX (FastMetal)
Mochi 1 previewApache License 2.0unknownComplete: calcuis GGUF Q3_K_M with Q4_0 T5 and FP8 VAE7.7 GB download#48 of 48 in text to video5,893 votessnapshot 2026-09-22Diffusers, ComfyUI
Wan2.2 TI2V 5BApache License 2.0unknownComplete: QuantStack GGUF Q8_013.5 GB downloadNo independent score yetBlaizzy/mlx-video, FastVideo MLX (FastMetal), ComfyUI
MiniMax H3MiniMax H3 Community License AgreementDoes not fit est.Complete: Unsloth GGUF Q4_K denoiser with the Q4_K_M Qwen3-VL encoder35.5 GB download#8 of 48 in text to video11,417 votessnapshot 2026-09-22Argus-AiTeam/minimax-h3-mac (MLX streaming), antirez/h3.c (Metal), ddalcu/mlx-serve, stable-diffusion.cpp, ComfyUI
Wan2.2 A14BApache License 2.0Does not fit est.Complete: QuantStack GGUF Q4_K_M (high and low noise)26.3 GB download#42 of 48 in text to video10,458 votessnapshot 2026-09-22PocketAI Wan2.2 MLX (pocketai-wan-video), ComfyUI

Quality: blind human votes from Arena (arena.ai), CC BY 4.0, snapshot 2026-09-22. Scores are for the Arena checkpoint, not your local quantized build.

24 GB Mac unified memory4 run · 1 unknown · 1 no

Best video models for a 24 GB Mac, with the reason for each verdict

Local models for this machine
ModelFitsQualityApp
LTX-2.3LTX-2 Community License AgreementRuns · PublishedApple M5 Pro · 24 GB unified memory · ltx-2-mlx-swift int8 streamed, 704x512, 121 frames, 150 s · ltx-2-mlx-swift READMENo independent score yetLightricks/LTX-2 ltx-pipelines, appautomaton/ltx-video-mlx, ComfyUI, dgrauet/ltx-2-mlx
MiniMax H3MiniMax H3 Community License AgreementRuns · PublishedComplete: Argus calibrated INT8 DiT with the official text encoder and VAEs115.5 GB download#8 of 48 in text to video11,417 votessnapshot 2026-09-22Argus-AiTeam/minimax-h3-mac (MLX streaming), antirez/h3.c (Metal), ddalcu/mlx-serve, stable-diffusion.cpp, ComfyUI
Wan2.1Apache License 2.0Runs · Stated16 GB unified memory · mlx-video 4-bit conversion · mlx-video Wan READMENo independent score yetml-explore/mlx-examples Wan2.1, ComfyUI, FastVideo MLX (FastMetal)
Wan2.2 TI2V 5BApache License 2.0Runs · PublishedComplete: FastMetal 5B QAD (DMD2-distilled INT8 derivative, not the base model)19.5 GB downloadNo independent score yetBlaizzy/mlx-video, FastVideo MLX (FastMetal), ComfyUI
Mochi 1 previewApache License 2.0unknownComplete: calcuis GGUF Q3_K_M with Q4_0 T5 and FP8 VAE7.7 GB download#48 of 48 in text to video5,893 votessnapshot 2026-09-22Diffusers, ComfyUI
Wan2.2 A14BApache License 2.0Does not fit est.Complete: QuantStack GGUF Q4_K_M (high and low noise)26.3 GB download#42 of 48 in text to video10,458 votessnapshot 2026-09-22PocketAI Wan2.2 MLX (pocketai-wan-video), ComfyUI

Quality: blind human votes from Arena (arena.ai), CC BY 4.0, snapshot 2026-09-22. Scores are for the Arena checkpoint, not your local quantized build.

32 GB Mac unified memory5 run · 2 unknown

Best video models for a 32 GB Mac, with the reason for each verdict

Local models for this machine
ModelFitsQualityApp
LTX-2.3LTX-2 Community License AgreementRuns · PublishedApple M2 Pro · 32 GB unified memory · ltx-2-mlx bf16 distilled with block streaming, 480x704, 33 frames · ltx-2-mlx project notesNo independent score yetLightricks/LTX-2 ltx-pipelines, appautomaton/ltx-video-mlx, ComfyUI, dgrauet/ltx-2-mlx
LTX-2.5LTX-2.x Community License AgreementRuns · PublishedApple M2 Pro · 32 GB unified memory · ltx-2-mlx q8 with --low-ram, 768x512, 49 frames, 277 s · ltx-2-mlx pipelines docsNo independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
MiniMax H3MiniMax H3 Community License AgreementRuns · Published on M4 Pro · 24 GBComplete: Argus calibrated INT8 DiT with the official text encoder and VAEs115.5 GB download#8 of 48 in text to video11,417 votessnapshot 2026-09-22Argus-AiTeam/minimax-h3-mac (MLX streaming), antirez/h3.c (Metal), ddalcu/mlx-serve, stable-diffusion.cpp, ComfyUI
Wan2.1Apache License 2.0Runs · Stated32 GB unified memory · Draw Things, 720p up to 81 frames · Draw Things wiki: video generationNo independent score yetml-explore/mlx-examples Wan2.1, ComfyUI, FastVideo MLX (FastMetal)
Wan2.2 TI2V 5BApache License 2.0Runs · Published on M5 · 24 GBComplete: FastMetal 5B QAD (DMD2-distilled INT8 derivative, not the base model)19.5 GB downloadNo independent score yetBlaizzy/mlx-video, FastVideo MLX (FastMetal), ComfyUI
Mochi 1 previewApache License 2.0unknownComplete: calcuis GGUF Q3_K_M with Q4_0 T5 and FP8 VAE7.7 GB download#48 of 48 in text to video5,893 votessnapshot 2026-09-22Diffusers, ComfyUI
Wan2.2 A14BApache License 2.0unknownComplete: PocketAI MLX Q4 (group size 64)28.6 GB download#42 of 48 in text to video10,458 votessnapshot 2026-09-22PocketAI Wan2.2 MLX (pocketai-wan-video), ComfyUI

Quality: blind human votes from Arena (arena.ai), CC BY 4.0, snapshot 2026-09-22. Scores are for the Arena checkpoint, not your local quantized build.

48 GB Mac unified memory5 run · 2 unknown

Best video models for a 32 GB Mac, with the reason for each verdict

Local models for this machine
ModelFitsQualityApp
LTX-2.3LTX-2 Community License AgreementRuns · PublishedComplete: dgrauet MLX int4 pack (distilled 1.1) with Gemma 3 12B 4-bit28.5 GB downloadNo independent score yetLightricks/LTX-2 ltx-pipelines, appautomaton/ltx-video-mlx, ComfyUI, dgrauet/ltx-2-mlx
LTX-2.5LTX-2.x Community License AgreementRuns · Stated48 GB unified memory · LTX Desktop on macOS 26+, full local generation · LTX Desktop README (Lightricks)No independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
MiniMax H3MiniMax H3 Community License AgreementRuns · PublishedComplete: Unsloth GGUF Q4_K denoiser with the Q4_K_M Qwen3-VL encoder35.5 GB download#8 of 48 in text to video11,417 votessnapshot 2026-09-22Argus-AiTeam/minimax-h3-mac (MLX streaming), antirez/h3.c (Metal), ddalcu/mlx-serve, stable-diffusion.cpp, ComfyUI
Wan2.1Apache License 2.0Runs · Stated32 GB unified memory · Draw Things, 720p up to 81 frames · Draw Things wiki: video generationNo independent score yetml-explore/mlx-examples Wan2.1, ComfyUI, FastVideo MLX (FastMetal)
Wan2.2 TI2V 5BApache License 2.0Runs · Published on M4 Max · 36 GBComplete: FastMetal 5B QAD (DMD2-distilled INT8 derivative, not the base model)19.5 GB downloadNo independent score yetBlaizzy/mlx-video, FastVideo MLX (FastMetal), ComfyUI
Mochi 1 previewApache License 2.0unknownComplete: calcuis GGUF Q3_K_M with Q4_0 T5 and FP8 VAE7.7 GB download#48 of 48 in text to video5,893 votessnapshot 2026-09-22Diffusers, ComfyUI
Wan2.2 A14BApache License 2.0unknownComplete: PocketAI MLX Q4 (group size 64)28.6 GB download#42 of 48 in text to video10,458 votessnapshot 2026-09-22PocketAI Wan2.2 MLX (pocketai-wan-video), ComfyUI

Quality: blind human votes from Arena (arena.ai), CC BY 4.0, snapshot 2026-09-22. Scores are for the Arena checkpoint, not your local quantized build.

64 GB Mac unified memory8 run

Best video models for a 64 GB Mac, with the reason for each verdict

Local models for this machine
ModelFitsQualityApp
HunyuanVideo-1.5Tencent Hunyuan Community LicenseRuns · PublishedApple M2 Ultra · 64 GB unified memory · ComfyUI official workflow, I2V 832x480, 73 frames · ComfyUI discussion #15109#38 of 48 in text to video4,292 votessnapshot 2026-09-22ComfyUI
LTX-2.3LTX-2 Community License AgreementRuns · Published on M5 Pro · 48 GBComplete: dgrauet MLX int4 pack (distilled 1.1) with Gemma 3 12B 4-bit28.5 GB downloadNo independent score yetLightricks/LTX-2 ltx-pipelines, appautomaton/ltx-video-mlx, ComfyUI, dgrauet/ltx-2-mlx
LTX-2.5LTX-2.x Community License AgreementRuns · PublishedApple M1 Max · 64 GB unified memory · KinoMLX FP16, 768x448, 5 s clip in 1 min 45 s · KinoMLX LTX-2.5 performanceNo independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
MiniMax H3MiniMax H3 Community License AgreementRuns · Published on M5 Pro · 48 GBComplete: Unsloth GGUF Q4_K denoiser with the Q4_K_M Qwen3-VL encoder35.5 GB download#8 of 48 in text to video11,417 votessnapshot 2026-09-22Argus-AiTeam/minimax-h3-mac (MLX streaming), antirez/h3.c (Metal), ddalcu/mlx-serve, stable-diffusion.cpp, ComfyUI
Mochi 1 previewApache License 2.0Runs · PublishedComplete: Official Diffusers BF16 variant40.0 GB download#48 of 48 in text to video5,893 votessnapshot 2026-09-22Diffusers, ComfyUI
Wan2.1Apache License 2.0Runs · Stated32 GB unified memory · Draw Things, 720p up to 81 frames · Draw Things wiki: video generationNo independent score yetml-explore/mlx-examples Wan2.1, ComfyUI, FastVideo MLX (FastMetal)
Wan2.2 A14BApache License 2.0Runs · PublishedApple M1 Max · 64 GB unified memory · ComfyUI Q5_K_M I2V with the Lightning 4-step LoRA, 480x320, 25 frames, 182.7 s · umamon on Zenn#42 of 48 in text to video10,458 votessnapshot 2026-09-22PocketAI Wan2.2 MLX (pocketai-wan-video), ComfyUI
Wan2.2 TI2V 5BApache License 2.0Runs · Published on M4 Max · 36 GBComplete: FastMetal 5B QAD (DMD2-distilled INT8 derivative, not the base model)19.5 GB downloadNo independent score yetBlaizzy/mlx-video, FastVideo MLX (FastMetal), ComfyUI

Quality: blind human votes from Arena (arena.ai), CC BY 4.0, snapshot 2026-09-22. Scores are for the Arena checkpoint, not your local quantized build.

96 GB Mac unified memory8 run

Best video models for a 64 GB Mac, with the reason for each verdict

Local models for this machine
ModelFitsQualityApp
HunyuanVideo-1.5Tencent Hunyuan Community LicenseRuns · PublishedApple M2 Ultra · 64 GB unified memory · ComfyUI official workflow, I2V 832x480, 73 frames · ComfyUI discussion #15109#38 of 48 in text to video4,292 votessnapshot 2026-09-22ComfyUI
LTX-2.3LTX-2 Community License AgreementRuns · Published on M5 Pro · 48 GBComplete: dgrauet MLX int4 pack (distilled 1.1) with Gemma 3 12B 4-bit28.5 GB downloadNo independent score yetLightricks/LTX-2 ltx-pipelines, appautomaton/ltx-video-mlx, ComfyUI, dgrauet/ltx-2-mlx
LTX-2.5LTX-2.x Community License AgreementRuns · PublishedApple M1 Max · 64 GB unified memory · KinoMLX FP16, 768x448, 5 s clip in 1 min 45 s · KinoMLX LTX-2.5 performanceNo independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
MiniMax H3MiniMax H3 Community License AgreementRuns · Published on M5 Pro · 48 GBComplete: Unsloth GGUF Q4_K denoiser with the Q4_K_M Qwen3-VL encoder35.5 GB download#8 of 48 in text to video11,417 votessnapshot 2026-09-22Argus-AiTeam/minimax-h3-mac (MLX streaming), antirez/h3.c (Metal), ddalcu/mlx-serve, stable-diffusion.cpp, ComfyUI
Mochi 1 previewApache License 2.0Runs · Published on M4 Max · 64 GBComplete: Official Diffusers BF16 variant40.0 GB download#48 of 48 in text to video5,893 votessnapshot 2026-09-22Diffusers, ComfyUI
Wan2.1Apache License 2.0Runs · Stated32 GB unified memory · Draw Things, 720p up to 81 frames · Draw Things wiki: video generationNo independent score yetml-explore/mlx-examples Wan2.1, ComfyUI, FastVideo MLX (FastMetal)
Wan2.2 A14BApache License 2.0Runs · PublishedApple M1 Max · 64 GB unified memory · ComfyUI Q5_K_M I2V with the Lightning 4-step LoRA, 480x320, 25 frames, 182.7 s · umamon on Zenn#42 of 48 in text to video10,458 votessnapshot 2026-09-22PocketAI Wan2.2 MLX (pocketai-wan-video), ComfyUI
Wan2.2 TI2V 5BApache License 2.0Runs · Published on M4 Max · 36 GBComplete: FastMetal 5B QAD (DMD2-distilled INT8 derivative, not the base model)19.5 GB downloadNo independent score yetBlaizzy/mlx-video, FastVideo MLX (FastMetal), ComfyUI

Quality: blind human votes from Arena (arena.ai), CC BY 4.0, snapshot 2026-09-22. Scores are for the Arena checkpoint, not your local quantized build.

128 GB Mac unified memory8 run

Best video models for a 64 GB Mac, with the reason for each verdict

Local models for this machine
ModelFitsQualityApp
HunyuanVideo-1.5Tencent Hunyuan Community LicenseRuns · PublishedApple M2 Ultra · 64 GB unified memory · ComfyUI official workflow, I2V 832x480, 73 frames · ComfyUI discussion #15109#38 of 48 in text to video4,292 votessnapshot 2026-09-22ComfyUI
LTX-2.3LTX-2 Community License AgreementRuns · PublishedComplete: Official FP8 dev checkpoint, quantized to 4-bit at load61.1 GB downloadNo independent score yetLightricks/LTX-2 ltx-pipelines, appautomaton/ltx-video-mlx, ComfyUI, dgrauet/ltx-2-mlx
LTX-2.5LTX-2.x Community License AgreementRuns · PublishedApple M1 Max · 64 GB unified memory · KinoMLX FP16, 768x448, 5 s clip in 1 min 45 s · KinoMLX LTX-2.5 performanceNo independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
MiniMax H3MiniMax H3 Community License AgreementRuns · PublishedComplete: MLX Serve FL2VA 4-bit pack with Turbo LoRA41.1 GB download#8 of 48 in text to video11,417 votessnapshot 2026-09-22Argus-AiTeam/minimax-h3-mac (MLX streaming), antirez/h3.c (Metal), ddalcu/mlx-serve, stable-diffusion.cpp, ComfyUI
Mochi 1 previewApache License 2.0Runs · Published on M4 Max · 64 GBComplete: Official Diffusers BF16 variant40.0 GB download#48 of 48 in text to video5,893 votessnapshot 2026-09-22Diffusers, ComfyUI
Wan2.1Apache License 2.0Runs · Stated32 GB unified memory · Draw Things, 720p up to 81 frames · Draw Things wiki: video generationNo independent score yetml-explore/mlx-examples Wan2.1, ComfyUI, FastVideo MLX (FastMetal)
Wan2.2 A14BApache License 2.0Runs · PublishedComplete: PocketAI MLX Q4 (group size 64)28.6 GB download#42 of 48 in text to video10,458 votessnapshot 2026-09-22PocketAI Wan2.2 MLX (pocketai-wan-video), ComfyUI
Wan2.2 TI2V 5BApache License 2.0Runs · Published on M4 Max · 36 GBComplete: FastMetal 5B QAD (DMD2-distilled INT8 derivative, not the base model)19.5 GB downloadNo independent score yetBlaizzy/mlx-video, FastVideo MLX (FastMetal), ComfyUI

Quality: blind human votes from Arena (arena.ai), CC BY 4.0, snapshot 2026-09-22. Scores are for the Arena checkpoint, not your local quantized build.

NVIDIA VRAM
8 GB NVIDIA VRAM4 run · 1 unknown

Best video models for 8 GB of VRAM, with the reason for each verdict

Local models for this machine
ModelFitsQualityApp
LTX-2.3LTX-2 Community License AgreementRuns · PublishedRTX 4060 · 8 GB VRAM · ComfyUI GGUF workflow, 720p 5 s in about 190 s, 32 GB RAM · ComfyUI issue #14345No independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
MiniMax H3MiniMax H3 Community License AgreementRuns · PublishedRTX 4050 · 6 GB VRAM · ComfyUI int8 convrot, 0.4 megapixel, 5 s, 16 GB RAM · ComfyUI issue #15482#8 of 48 in text to video11,417 votessnapshot 2026-09-22stable-diffusion.cpp, ComfyUI
Wan2.2 A14BApache License 2.0Runs · PublishedRTX 2060 · 6 GB VRAM · ComfyUI Q8 GGUF, 480x720, 81 frames, 7 to 10 min, 28 GB RAM plus a 60 GB page file · ComfyUI issue #12451#42 of 48 in text to video10,458 votessnapshot 2026-09-22ComfyUI
Wan2.2 TI2V 5BApache License 2.0Runs · Stated8 GB VRAM · ComfyUI native offloading · ComfyUI docs: Wan2.2Complete: Comfy-Org repackaged FP16 with FP8 scaled UMT518.1 GB downloadNo independent score yetComfyUI
Mochi 1 previewApache License 2.0unknownComplete: calcuis GGUF Q3_K_M with Q4_0 T5 and FP8 VAE7.7 GB download#48 of 48 in text to video5,893 votessnapshot 2026-09-22Diffusers, ComfyUI

Quality: blind human votes from Arena (arena.ai), CC BY 4.0, snapshot 2026-09-22. Scores are for the Arena checkpoint, not your local quantized build.

12 GB NVIDIA VRAM6 run · 1 unknown

Best video models for 12 GB of VRAM, with the reason for each verdict

Local models for this machine
ModelFitsQualityApp
HunyuanVideo-1.5Tencent Hunyuan Community LicenseRuns · PublishedRTX 4070 · 12 GB VRAM · ComfyUI example workflow, 480p I2V distilled fp8, 81 frames · ComfyUI issue #10855#38 of 48 in text to video4,292 votessnapshot 2026-09-22ComfyUI
LTX-2.3LTX-2 Community License AgreementRuns · PublishedRTX 4060 · 8 GB VRAM · ComfyUI GGUF workflow, 720p 5 s in about 190 s, 32 GB RAM · ComfyUI issue #14345No independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
LTX-2.5LTX-2.x Community License AgreementRuns · Stated12 GB VRAM · LTX Desktop on Windows, weight streaming, 40 GB RAM, 540p · LTX Desktop compatibility matrix (Lightricks)No independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
MiniMax H3MiniMax H3 Community License AgreementRuns · Stated12 GB VRAM · Diffusers int8, streamed blocks, 960x544, about 75 GB system RAM · Diffusers docs: MiniMax-H3 memory#8 of 48 in text to video11,417 votessnapshot 2026-09-22stable-diffusion.cpp, ComfyUI
Wan2.2 A14BApache License 2.0Runs · PublishedRTX 2060 · 6 GB VRAM · ComfyUI Q8 GGUF, 480x720, 81 frames, 7 to 10 min, 28 GB RAM plus a 60 GB page file · ComfyUI issue #12451#42 of 48 in text to video10,458 votessnapshot 2026-09-22ComfyUI
Wan2.2 TI2V 5BApache License 2.0Runs · Stated8 GB VRAM · ComfyUI native offloading · ComfyUI docs: Wan2.2Complete: Comfy-Org repackaged FP16 with FP8 scaled UMT518.1 GB downloadNo independent score yetComfyUI
Mochi 1 previewApache License 2.0unknownComplete: calcuis GGUF Q3_K_M with Q4_0 T5 and FP8 VAE7.7 GB download#48 of 48 in text to video5,893 votessnapshot 2026-09-22Diffusers, ComfyUI

Quality: blind human votes from Arena (arena.ai), CC BY 4.0, snapshot 2026-09-22. Scores are for the Arena checkpoint, not your local quantized build.

16 GB NVIDIA VRAM7 run · 1 unknown

Best video models for 16 GB of VRAM, with the reason for each verdict

Local models for this machine
ModelFitsQualityApp
HunyuanVideo-1.5Tencent Hunyuan Community LicenseRuns · Stated14 GB VRAM · official runner with model offloading · HunyuanVideo-1.5 model card#38 of 48 in text to video4,292 votessnapshot 2026-09-22ComfyUI
LTX-2.3LTX-2 Community License AgreementRuns · Stated16 GB VRAM · distilled with FP8 cast, previews only · LTX blog: hardware for AI video modelsNo independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
LTX-2.5LTX-2.x Community License AgreementRuns · Stated12 GB VRAM · LTX Desktop on Windows, weight streaming, 40 GB RAM, 540p · LTX Desktop compatibility matrix (Lightricks)No independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
MiniMax H3MiniMax H3 Community License AgreementRuns · PublishedRTX 5060 · 16 GB VRAM · ComfyUI int8 convrot with int4 text encoder, 1024x768, 10 s in 24 min · ComfyUI issue #15274#8 of 48 in text to video11,417 votessnapshot 2026-09-22stable-diffusion.cpp, ComfyUI
Wan2.1Apache License 2.0Runs · Stated13 GB VRAM · T2V 14B in Diffusers, bf16 with group offloading to system RAM · Diffusers docs: Wan pipelineNo independent score yetComfyUI
Wan2.2 A14BApache License 2.0Runs · PublishedRTX 5060 Ti · 16 GB VRAM · ComfyUI fp16 I2V with 4-step LoRAs, 64 GB system RAM · ComfyUI issue #10662#42 of 48 in text to video10,458 votessnapshot 2026-09-22ComfyUI
Wan2.2 TI2V 5BApache License 2.0Runs · Stated8 GB VRAM · ComfyUI native offloading · ComfyUI docs: Wan2.2Complete: Comfy-Org repackaged FP16 with FP8 scaled UMT518.1 GB downloadNo independent score yetComfyUI
Mochi 1 previewApache License 2.0unknownComplete: calcuis GGUF Q3_K_M with Q4_0 T5 and FP8 VAE7.7 GB download#48 of 48 in text to video5,893 votessnapshot 2026-09-22Diffusers, ComfyUI

Quality: blind human votes from Arena (arena.ai), CC BY 4.0, snapshot 2026-09-22. Scores are for the Arena checkpoint, not your local quantized build.

24 GB NVIDIA VRAM8 run

Best video models for 24 GB of VRAM, with the reason for each verdict

Local models for this machine
ModelFitsQualityApp
HunyuanVideo-1.5Tencent Hunyuan Community LicenseRuns · Stated24 GB VRAM · ComfyUI native workflow · ComfyUI docs: HunyuanVideo 1.5#38 of 48 in text to video4,292 votessnapshot 2026-09-22ComfyUI
LTX-2.3LTX-2 Community License AgreementRuns · Stated24 GB VRAM · two-stage production at 720p to 1080p · LTX blog: hardware for AI video modelsNo independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
LTX-2.5LTX-2.x Community License AgreementRuns · Stated24 GB VRAM · LTX Desktop app, 32 GB system RAM · LTX Desktop README (Lightricks)No independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
MiniMax H3MiniMax H3 Community License AgreementRuns · Stated24 GB VRAM · Diffusers int8 with streamed blocks, about 75 GB system RAM · Diffusers docs: MiniMax-H3 memory#8 of 48 in text to video11,417 votessnapshot 2026-09-22stable-diffusion.cpp, ComfyUI
Mochi 1 previewApache License 2.0Runs · Stated22 GB VRAM · Diffusers bf16 variant with CPU offload · Mochi 1 model cardComplete: Official Diffusers BF16 variant40.0 GB download#48 of 48 in text to video5,893 votessnapshot 2026-09-22Diffusers, ComfyUI
Wan2.1Apache License 2.0Runs · Stated13 GB VRAM · T2V 14B in Diffusers, bf16 with group offloading to system RAM · Diffusers docs: Wan pipelineNo independent score yetComfyUI
Wan2.2 A14BApache License 2.0Runs · PublishedRTX 5060 Ti · 16 GB VRAM · ComfyUI fp16 I2V with 4-step LoRAs, 64 GB system RAM · ComfyUI issue #10662#42 of 48 in text to video10,458 votessnapshot 2026-09-22ComfyUI
Wan2.2 TI2V 5BApache License 2.0Runs · Stated24 GB VRAM · official generate.py at 1280x704 with offload flags · Wan2.2-TI2V-5B model cardNo independent score yetComfyUI

Quality: blind human votes from Arena (arena.ai), CC BY 4.0, snapshot 2026-09-22. Scores are for the Arena checkpoint, not your local quantized build.

32 GB NVIDIA VRAM8 run

Best video models for 24 GB of VRAM, with the reason for each verdict

Local models for this machine
ModelFitsQualityApp
HunyuanVideo-1.5Tencent Hunyuan Community LicenseRuns · Stated24 GB VRAM · ComfyUI native workflow · ComfyUI docs: HunyuanVideo 1.5#38 of 48 in text to video4,292 votessnapshot 2026-09-22ComfyUI
LTX-2.3LTX-2 Community License AgreementRuns · Stated24 GB VRAM · two-stage production at 720p to 1080p · LTX blog: hardware for AI video modelsNo independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
LTX-2.5LTX-2.x Community License AgreementRuns · Stated32 GB VRAM · official requirement, plus 32 GB system RAM · LTX-2.5 page (ltx.io)No independent score yetLightricks/LTX-2 ltx-pipelines, ComfyUI
MiniMax H3MiniMax H3 Community License AgreementRuns · Stated24 GB VRAM · Diffusers int8 with streamed blocks, about 75 GB system RAM · Diffusers docs: MiniMax-H3 memory#8 of 48 in text to video11,417 votessnapshot 2026-09-22stable-diffusion.cpp, ComfyUI
Mochi 1 previewApache License 2.0Runs · Stated22 GB VRAM · Diffusers bf16 variant with CPU offload · Mochi 1 model cardComplete: Official Diffusers BF16 variant40.0 GB download#48 of 48 in text to video5,893 votessnapshot 2026-09-22Diffusers, ComfyUI
Wan2.1Apache License 2.0Runs · Stated13 GB VRAM · T2V 14B in Diffusers, bf16 with group offloading to system RAM · Diffusers docs: Wan pipelineNo independent score yetComfyUI
Wan2.2 A14BApache License 2.0Runs · PublishedRTX 5060 Ti · 16 GB VRAM · ComfyUI fp16 I2V with 4-step LoRAs, 64 GB system RAM · ComfyUI issue #10662#42 of 48 in text to video10,458 votessnapshot 2026-09-22ComfyUI
Wan2.2 TI2V 5BApache License 2.0Runs · Stated24 GB VRAM · official generate.py at 1280x704 with offload flags · Wan2.2-TI2V-5B model cardNo independent score yetComfyUI

Quality: blind human votes from Arena (arena.ai), CC BY 4.0, snapshot 2026-09-22. Scores are for the Arena checkpoint, not your local quantized build.

Frequently asked questions

Can a Mac generate AI videos locally?

Yes. Apple Silicon Macs can run open-weight video models locally. Choose your unified memory below to see compatible builds, Mac apps and any published generation times.

How much memory does a local AI video generator need?

It depends on the model, quantization and app. Macs use unified memory and NVIDIA GPUs use VRAM. ModelFit accounts for the complete component bundle and known runtime costs. When those costs are unknown, the fit stays unknown instead of treating download size as memory use.

How fast is local AI video generation on a Mac?

It varies by model, chip, memory, app and settings. The tables only show published measurements for the exact Mac, build and workload; otherwise Time is unknown. ModelFit does not run its own benchmarks.

Keep exploring