[{"data":1,"prerenderedAt":1375},["ShallowReactive",2],{"blog-post-en-\u002Fblog\u002Fpaiton-video-ltx-2-5-minimax-h3-radeon-r9700":3,"blog-posts-sidebar-en":785},{"id":4,"title":5,"body":6,"categories":765,"date":770,"description":771,"extension":772,"heading":5,"image":173,"meta":773,"navigation":166,"originalUrl":774,"path":775,"seo":776,"slug":777,"socialImage":173,"stem":778,"updated":779,"video":780,"__hash__":784},"blog\u002Fblog\u002Fpaiton-video-ltx-2-5-minimax-h3-radeon-r9700.md","Video generation on the Radeon R9700: LTX-2.5 and MiniMax H3",{"type":7,"value":8,"toc":757},"minimark",[9,44,47,110,113,119,122,156,161,175,178,182,185,193,198,273,276,279,282,291,297,302,306,314,317,324,329,435,438,441,444,447,455,461,471,475,484,487,496,585,588,600,607,685,688,691,700,709,713,753],[10,11,12,13,18,19,23,24,28,29,33,34,38,39,43],"p",{},"The Radeon AI PRO R9700 already does a lot of local AI work with ",[14,15,17],"a",{"href":16},"\u002Fproducts\u002Fpaiton","Paiton",".\n",[14,20,22],{"href":21},"\u002Fblog\u002Fpaiton-qwen38-w3a4-radeon-ai-pro-r9700","Qwen3.8 27B"," answers coding questions,\n",[14,25,27],{"href":26},"\u002Fblog\u002Fpaiton-flux2-klein-radeon-ai-pro-r9700","FLUX.2 klein"," draws an image in about a second,\n",[14,30,32],{"href":31},"\u002Fblog\u002Fpaiton-qwen-image-21-radeon-ai-pro-r9700","Qwen-Image 2.1"," and\n",[14,35,37],{"href":36},"\u002Fblog\u002Fpaiton-minimax-h3-image-radeon-ai-pro-r9700","MiniMax-H3-Image"," generate and edit images, and\n",[14,40,42],{"href":41},"\u002Fblog\u002Fpaiton-idol-radeon-ai-pro-r9700","Idol"," turns one picture into a 3D asset in about 16 seconds.",[10,45,46],{},"Video was the gap. Our earlier video packages ran inside ComfyUI and PyTorch, where Paiton could only speed up part of\nthe work: from no gain on Wan2.2 to about 15% on MiniMax H3. This week we released two video models with native Paiton\ninference from the prompt through the video and audio decoders. Both make the soundtrack together with the picture:",[48,49,50,67],"table",{},[51,52,53],"thead",{},[54,55,56,60,63],"tr",{},[57,58,59],"th",{},"Model",[57,61,62],{},"What it does",[57,64,66],{"align":65},"right","5-second clip on one R9700",[68,69,70,92],"tbody",{},[54,71,72,82,85],{},[73,74,75,81],"td",{},[14,76,80],{"href":77,"rel":78},"https:\u002F\u002Fgithub.com\u002FEliovp-BV\u002Fpaiton-vllm-plugin\u002Fblob\u002Fmain\u002Fmodels\u002FLTX-2.5\u002FREADME.md",[79],"nofollow","LTX-2.5",", Lightricks, 22B",[73,83,84],{},"Text or a first frame to video with audio, 704p and 1080p",[73,86,87,91],{"align":65},[88,89,90],"strong",{},"18.1 s"," at 1280 x 704; 48.8 s at 1920 x 1088",[54,93,94,101,104],{},[73,95,96],{},[14,97,100],{"href":98,"rel":99},"https:\u002F\u002Fgithub.com\u002FEliovp-BV\u002Fpaiton-vllm-plugin\u002Freleases\u002Ftag\u002Fminimax-h3-rdna4-v1.2.0",[79],"MiniMax H3 v1.2",[73,102,103],{},"Video with stereo audio from a prompt and up to nine reference images",[73,105,106,109],{"align":65},[88,107,108],{},"30.4 s"," with one reference; 44.3 s with nine, at 864 x 480",[10,111,112],{},"Both fit within the card's 32 GB of VRAM: the peaks we measured were 27.4 GiB for LTX-2.5 at 1080p and 26.8 GiB for\nMiniMax H3. That is GPU memory, not the whole system's memory use. Both also stream text-encoder layers from system RAM.\nThe timing boundaries differ between the two models and are explained below.",[10,114,115],{},[116,117,118],"em",{},"The cover is an unaltered frame from the AI-generated LTX-2.5 clip below.",[10,120,121],{},"In this article",[123,124,125,132,138,144,150],"ul",{},[126,127,128],"li",{},[14,129,131],{"href":130},"#watch-the-clip","Watch the LTX-2.5 clip",[126,133,134],{},[14,135,137],{"href":136},"#ltx-25-a-5-second-clip-in-18-seconds","LTX-2.5 results and quality",[126,139,140],{},[14,141,143],{"href":142},"#minimax-h3-v12-video-from-your-own-reference-images","MiniMax H3 v1.2 results and quality",[126,145,146],{},[14,147,149],{"href":148},"#try-it","Try it on your R9700",[126,151,152],{},[14,153,155],{"href":154},"#licenses","Licenses",[157,158,160],"h2",{"id":159},"watch-the-clip","Watch the clip",[10,162,163],{},[164,165],"video",{"controls":166,"playsInline":166,"preload":167,"width":168,"height":169,"ariaLabel":170,"ariaDescribedBy":171,"poster":173,"src":174},true,"none",1280,704,"AI-generated street drummer and crowd, with the model's original audio",[172],"ltx25-video-caption","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-video-r9700\u002Fltx25-drummer-704.jpg","\u002Fasset\u002Fvideos\u002Fblog\u002Fpaiton-video-r9700\u002Fltx25-drummer-704.mp4",[10,176,177],{"id":172},"AI-generated video and audio from LTX-2.5 on one R9700, unaltered from the supplied example. Prompt: \"A street drummer plays a fast rhythm on a city sidewalk at dusk while the crowd claps and cheers\". 1280 x 704, 121 frames at 24 fps, about 5 seconds of video. The measured warm request takes 18.1 seconds. Press play to hear the soundtrack.",[157,179,181],{"id":180},"ltx-25-a-5-second-clip-in-18-seconds","LTX-2.5: a 5-second clip in 18 seconds",[10,183,184],{},"LTX-2.5 is Lightricks' 22B audio and video model. We run its distilled two-stage pipeline: 8 steps at half size, a 2x\nupscale of the latent, then 3 steps at full size, followed by the video and audio decoders. On one R9700 the stock\npipeline cannot hold the 16-bit weights, so it has to offload them to the CPU, and a 1080p clip takes seven minutes.\nPaiton's native GPU pipeline brings that down to 48.8 seconds, while streaming the text-encoder layers that do not fit\nin VRAM from host memory.",[10,186,187],{},[188,189],"img",{"alt":190,"src":191,"png-fallback":192},"LTX-2.5 text-to-video time on one R9700. At 1280 x 704: stock BF16 128.6 seconds, Paiton 18.1 seconds. At 1920 x 1088: stock BF16 420.5 seconds, Paiton 48.8 seconds. Lower is better.","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-video-r9700\u002Fltx25.svg","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-video-r9700\u002Fltx25.png",[10,194,195],{},[116,196,197],{},"121 frames with audio, one R9700 at 300 W. Paiton: warm server-side HTTP request through finished MP4, median of three runs. Stock: warm BF16 pipeline with CPU offload. Paiton uses FP8, so this compares serving configurations, not identical precision.",[48,199,200,215],{},[51,201,202],{},[54,203,204,207,209,212],{},[57,205,206],{},"Request, 121 frames, about 5 s at 24 fps",[57,208,17],{"align":65},[57,210,211],{"align":65},"Stock LTX-2.5, BF16",[57,213,214],{"align":65},"Speedup",[68,216,217,232,245,261],{},[54,218,219,222,224,227],{},[73,220,221],{},"Text to video, 1280 x 704",[73,223,90],{"align":65},[73,225,226],{"align":65},"128.6 s",[73,228,229],{"align":65},[88,230,231],{},"7.1x",[54,233,234,237,240,243],{},[73,235,236],{},"Image to video, 1280 x 704",[73,238,239],{"align":65},"21.1 s",[73,241,242],{"align":65},"Not measured",[73,244,242],{"align":65},[54,246,247,250,253,256],{},[73,248,249],{},"Text to video, 1920 x 1088",[73,251,252],{"align":65},"48.8 s",[73,254,255],{"align":65},"420.5 s",[73,257,258],{"align":65},[88,259,260],{},"8.6x",[54,262,263,266,269,271],{},[73,264,265],{},"Image to video, 1920 x 1088",[73,267,268],{"align":65},"51.9 s",[73,270,242],{"align":65},[73,272,242],{"align":65},[10,274,275],{},"Paiton's times are measured server-side from the HTTP request to the finished MP4, warm, the median of three runs, with the card at its\n300 W limit. They include the text encoder, both sampling stages, the upscale, both decoders and the H.264 and AAC\nencoding. Startup and weight downloads are excluded.",[10,277,278],{},"Most of the speed comes from running the transformer in FP8, both weights and activations, and the attention in INT8\nand FP8, on kernels written for the R9700's RDNA4 chip. The timestep conditioning is computed once ahead of time for\nthe distilled schedule's nine fixed noise levels. The steps run as HIP graphs, and the 12B text encoder streams the\nlayers that do not fit on the card from host memory.",[10,280,281],{},"Quality matters here. An FP8 copy of this 11-step sampler does not reproduce the 16-bit model's video exactly: small\nnumeric differences can grow over the steps into a different camera path or layout. Lightricks' own FP8 mode does the\nsame. We compared both FP8 versions with the 16-bit run from the same noise. At 704p, Paiton is as close to it as the\nstock FP8 mode: 25.7 dB against 25.3 dB PSNR. At 1080p it is 1.1 dB further away on the mean of five prompts, with two\nprompts lower and three equal or higher. We looked at the frames side by side and found no blocking, banding or noise\nin those cases. This is a small quality check, not a promise of identical videos.",[10,283,284,285,290],{},"The ",[14,286,289],{"href":287,"rel":288},"https:\u002F\u002Fgithub.com\u002FEliovp-BV\u002Fpaiton-vllm-plugin\u002Fblob\u002Fmain\u002Fmodels\u002FLTX-2.5\u002FREADME.md#limits-of-this-build",[79],"LTX-2.5 guide","\nlists the full numbers and the limits of this first build: two measured sizes, no portrait sizes, and the clip length\nhas to be given.",[10,292,293],{},[188,294],{"alt":295,"src":296},"A golden dog running along the shore at sunset. An unaltered frame from the supplied AI-generated LTX-2.5 image-to-video example at 1920 x 1088.","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-video-r9700\u002Fltx25-i2v-1088.jpg",[10,298,299],{},[116,300,301],{},"AI-generated image-to-video example at 1920 x 1088. The measured request produces 121 frames with audio in 51.9 s.",[157,303,305],{"id":304},"minimax-h3-v12-video-from-your-own-reference-images","MiniMax H3 v1.2: video from your own reference images",[10,307,308,309,313],{},"MiniMax H3 makes a clip from a prompt plus references: up to nine images, or a video whose soundtrack it uses as\nreference audio. You can write \"the red panda from ",[310,311,312],"code",{},"\u003CPicture 1>"," walks through a sunlit forest\" to guide the clip with\nyour own image. The audio is 32 kHz stereo.",[10,315,316],{},"Our earlier v1.1.0 package used ComfyUI, with Paiton replacing the sampler. v1.2 replaces the other model components\ntoo: the 32B text encoder, the reference encoder, the transformer and the video and audio decoders now run on Paiton's\nHIP kernels, with no PyTorch, Triton or ComfyUI in the container. At 864 x 480, the gain over the stock engine grows\nwith the number of references:",[10,318,319],{},[188,320],{"alt":321,"src":322,"png-fallback":323},"MiniMax H3 time on one R9700, stock versus Paiton v1.2. At 864 x 480 with one, three and nine references: 67.4 versus 30.4, 91.2 versus 33.4, and 167.5 versus 44.3 seconds. At 1344 x 768 with nine references: 278.7 versus 114.1 seconds. Lower is better.","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-video-r9700\u002Fminimax-h3.svg","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-video-r9700\u002Fminimax-h3.png",[10,325,326],{},[116,327,328],{},"124 frames, Turbo4, one R9700 at 300 W. Stock: warm mean of two runs, reference loading through written MP4. Paiton v1.2: published M2 run, prepared inputs through written MP4, excluding CPU input preparation and weight loading. Stock uses W4A8; v1.2 uses FP8.",[48,330,331,350],{},[51,332,333],{},[54,334,335,338,341,344,347],{},[57,336,337],{},"Request, 124 frames, about 5 s, four-step Turbo4",[57,339,340],{"align":65},"Stock",[57,342,343],{"align":65},"Paiton v1.1.0",[57,345,346],{"align":65},"Paiton v1.2",[57,348,349],{"align":65},"v1.2 against stock",[68,351,352,368,385,404,421],{},[54,353,354,357,360,363,365],{},[73,355,356],{},"864 x 480, 1 image reference",[73,358,359],{"align":65},"67.4 s",[73,361,362],{"align":65},"50.6 s",[73,364,108],{"align":65},[73,366,367],{"align":65},"2.2x",[54,369,370,373,376,379,382],{},[73,371,372],{},"864 x 480, 3 image references",[73,374,375],{"align":65},"91.2 s",[73,377,378],{"align":65},"85.3 s",[73,380,381],{"align":65},"33.4 s",[73,383,384],{"align":65},"2.7x",[54,386,387,390,393,396,399],{},[73,388,389],{},"864 x 480, 9 image references",[73,391,392],{"align":65},"167.5 s",[73,394,395],{"align":65},"178.7 s",[73,397,398],{"align":65},"44.3 s",[73,400,401],{"align":65},[88,402,403],{},"3.8x",[54,405,406,409,412,415,418],{},[73,407,408],{},"1344 x 768, 9 image references",[73,410,411],{"align":65},"278.7 s",[73,413,414],{"align":65},"258.4 s",[73,416,417],{"align":65},"114.1 s",[73,419,420],{"align":65},"2.4x",[54,422,423,426,428,430,433],{},[73,424,425],{},"864 x 480, 1 video reference with its soundtrack",[73,427,242],{"align":65},[73,429,242],{"align":65},[73,431,432],{"align":65},"203.8 s",[73,434,242],{"align":65},[10,436,437],{},"The stock comparison supplied for this article uses the MiniMax H3 reference-to-video workflow in ComfyUI with its own\nengine, on the same card, with the same prompts, reference images, sizes and seed as v1.2. It runs the W4A8 checkpoint our v1.1.0 package ships, while v1.2 runs\nFP8 weights. The v1.1.0 column comes from the October release, with other prompts and images of the same sizes; it is\nhistorical context, not a matched three-way test.",[10,439,440],{},"The v1.2 measurements start with prepared inputs and end with the written MP4. They exclude CPU input preparation,\nreported as less than two seconds, and weight loading. With many references, the stock engine spends more than half\nits time encoding them: 89 seconds of the 168 seconds with nine. That is where v1.2 gains the most.",[10,442,443],{},"The eight-step Turbo8 preset takes 51.4 s for the first row in v1.2. Peak VRAM is under 27 GiB in v1.2, against 30.5 to\n30.8 GiB reported for v1.1.0.",[10,445,446],{},"For quality, we compared five cases with the 16-bit model on an Instinct MI355X, using the same prompts, references,\nseed and starting noise. Prompt alignment, measured by CLIP score, is the same as the 16-bit model's within the spread\nbetween cases. With one or three references, the clips stay close to the 16-bit ones. With nine references, the\nsubjects and the scene are all there, but they can stand in different places.",[10,448,284,449,454],{},[14,450,453],{"href":451,"rel":452},"https:\u002F\u002Fgithub.com\u002FEliovp-BV\u002Fpaiton-vllm-plugin\u002Fblob\u002Fmain\u002Fmodels\u002FMiniMax-H3\u002FBENCHMARKS.md#v12-native-runner-release-candidate",[79],"benchmark report","\nhas the v1.2 per-case scores and limits: up to about 9.4 seconds at 768p with nine references, no audio-only references,\nand Base50 stays on v1.1.0.",[10,456,457],{},[188,458],{"alt":459,"src":460},"A fox beside a stream. Unaltered frame from the earlier 15-second AI-generated MiniMax H3 example on one R9700.","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-video-r9700\u002Fminimax-h3-fox-15s-poster.jpg",[10,462,463],{},[116,464,465,466,470],{},"An earlier MiniMax H3 example, not a v1.2 benchmark output. The full 15-second clip is in our ",[14,467,469],{"href":468},"\u002Fblog\u002Fpaiton-minimax-h3-radeon-ai-pro-r9700","original video article",".",[157,472,474],{"id":473},"try-it","Try it",[10,476,477,478,483],{},"Both models start with one command from a checkout of the\n",[14,479,482],{"href":480,"rel":481},"https:\u002F\u002Fgithub.com\u002FEliovp-BV\u002Fpaiton-vllm-plugin",[79],"paiton-vllm-plugin"," repository. You need one R9700, Linux, Docker and a\nworking ROCm kernel driver. The first start downloads the weights, about 38 GB for LTX-2.5 or 51 GB for MiniMax H3\nTurbo4, and checks every file against its SHA-256.",[10,485,486],{},"Allow for system RAM as well as VRAM. LTX-2.5 pins about 14.1 GB of text-encoder weights in host memory; MiniMax H3's\nfastest text-encoder mode needs about 41 GB of free host RAM and has not been tested on hosts below 64 GB. The guides\ncover the full RAM and disk requirements.",[10,488,489,490,495],{},"For LTX-2.5, accept the license on the ",[14,491,494],{"href":492,"rel":493},"https:\u002F\u002Fhuggingface.co\u002FEliovpAI\u002FLTX-2.5-FP8-Paiton-RDNA4",[79],"weights page"," first.\nReplace the token placeholder below with a Hugging Face read token from that same account:",[497,498,503],"pre",{"className":499,"code":500,"language":501,"meta":502,"style":502},"language-bash shiki shiki-themes github-light github-dark","git clone --depth 1 https:\u002F\u002Fgithub.com\u002FEliovp-BV\u002Fpaiton-vllm-plugin.git\ncd paiton-vllm-plugin\nexport HF_TOKEN='hf_your_read_token'\n.\u002Fmodels\u002FLTX-2.5\u002Flaunch.sh\n# In a second terminal, from the same checkout, once it prints READY:\n.\u002Fmodels\u002FLTX-2.5\u002Frun.sh generate \"A street drummer plays a fast rhythm on a city sidewalk at dusk while the crowd claps and cheers\" --out drummer.mp4\n","bash","",[310,504,505,528,537,554,560,567],{"__ignoreMap":502},[506,507,510,514,518,522,525],"span",{"class":508,"line":509},"line",1,[506,511,513],{"class":512},"sScJk","git",[506,515,517],{"class":516},"sZZnC"," clone",[506,519,521],{"class":520},"sj4cs"," --depth",[506,523,524],{"class":520}," 1",[506,526,527],{"class":516}," https:\u002F\u002Fgithub.com\u002FEliovp-BV\u002Fpaiton-vllm-plugin.git\n",[506,529,531,534],{"class":508,"line":530},2,[506,532,533],{"class":520},"cd",[506,535,536],{"class":516}," paiton-vllm-plugin\n",[506,538,540,544,548,551],{"class":508,"line":539},3,[506,541,543],{"class":542},"szBVR","export",[506,545,547],{"class":546},"sVt8B"," HF_TOKEN",[506,549,550],{"class":542},"=",[506,552,553],{"class":516},"'hf_your_read_token'\n",[506,555,557],{"class":508,"line":556},4,[506,558,559],{"class":512},".\u002Fmodels\u002FLTX-2.5\u002Flaunch.sh\n",[506,561,563],{"class":508,"line":562},5,[506,564,566],{"class":565},"sJ8bj","# In a second terminal, from the same checkout, once it prints READY:\n",[506,568,570,573,576,579,582],{"class":508,"line":569},6,[506,571,572],{"class":512},".\u002Fmodels\u002FLTX-2.5\u002Frun.sh",[506,574,575],{"class":516}," generate",[506,577,578],{"class":516}," \"A street drummer plays a fast rhythm on a city sidewalk at dusk while the crowd claps and cheers\"",[506,580,581],{"class":520}," --out",[506,583,584],{"class":516}," drummer.mp4\n",[10,586,587],{},"Both servers default to port 30000. Stop LTX-2.5 before trying MiniMax H3 on that port:",[497,589,591],{"className":499,"code":590,"language":501,"meta":502,"style":502},".\u002Fmodels\u002FLTX-2.5\u002Frun.sh stop\n",[310,592,593],{"__ignoreMap":502},[506,594,595,597],{"class":508,"line":509},[506,596,572],{"class":512},[506,598,599],{"class":516}," stop\n",[10,601,602,603,606],{},"For MiniMax H3 v1.2, from the same checkout, with a reference image saved as ",[310,604,605],{},"panda.png",":",[497,608,610],{"className":499,"code":609,"language":501,"meta":502,"style":502},"PAITON_DETACH=1 .\u002Fmodels\u002FMiniMax-H3\u002Fnative\u002Flaunch.sh http\ndocker logs -f paiton-minimax-h3-v12\n# Wait for READY, then run the request in another terminal:\ncurl --fail-with-body -o panda.mp4 -F image=@panda.png \\\n  -F prompt='The red panda from \u003CPicture 1> walks through a sunlit forest. Audio: birds and a stream.' \\\n  http:\u002F\u002F127.0.0.1:30000\u002Fv1\u002Fvideos\u002Fgenerations\n",[310,611,612,628,642,647,670,680],{"__ignoreMap":502},[506,613,614,617,619,622,625],{"class":508,"line":509},[506,615,616],{"class":546},"PAITON_DETACH",[506,618,550],{"class":542},[506,620,621],{"class":516},"1",[506,623,624],{"class":512}," .\u002Fmodels\u002FMiniMax-H3\u002Fnative\u002Flaunch.sh",[506,626,627],{"class":516}," http\n",[506,629,630,633,636,639],{"class":508,"line":530},[506,631,632],{"class":512},"docker",[506,634,635],{"class":516}," logs",[506,637,638],{"class":520}," -f",[506,640,641],{"class":516}," paiton-minimax-h3-v12\n",[506,643,644],{"class":508,"line":539},[506,645,646],{"class":565},"# Wait for READY, then run the request in another terminal:\n",[506,648,649,652,655,658,661,664,667],{"class":508,"line":556},[506,650,651],{"class":512},"curl",[506,653,654],{"class":520}," --fail-with-body",[506,656,657],{"class":520}," -o",[506,659,660],{"class":516}," panda.mp4",[506,662,663],{"class":520}," -F",[506,665,666],{"class":516}," image=@panda.png",[506,668,669],{"class":520}," \\\n",[506,671,672,675,678],{"class":508,"line":562},[506,673,674],{"class":520},"  -F",[506,676,677],{"class":516}," prompt='The red panda from \u003CPicture 1> walks through a sunlit forest. Audio: birds and a stream.'",[506,679,669],{"class":520},[506,681,682],{"class":508,"line":569},[506,683,684],{"class":516},"  http:\u002F\u002F127.0.0.1:30000\u002Fv1\u002Fvideos\u002Fgenerations\n",[10,686,687],{},"Both serve an HTTP API. H3 also comes with a ComfyUI node that sends requests to the native server. These are\nmodel-specific video launchers, separate from the language-model presets.",[157,689,155],{"id":690},"licenses",[10,692,693,694,699],{},"The repository code is Apache-2.0, but that does not replace the model licenses or the terms for derivative runtimes.\nLTX-2.5 uses the ",[14,695,698],{"href":696,"rel":697},"https:\u002F\u002Fgithub.com\u002FEliovp-BV\u002Fpaiton-vllm-plugin\u002Fblob\u002Fmain\u002Fmodels\u002FLTX-2.5\u002FLICENSES\u002FLTX-2.x-Community-License.txt",[79],"LTX-2.x Community License",".\nCompanies with US$10 million or more in annual revenue need a paid license from Lightricks for use beyond\nnon-commercial testing and research. Published generated media must be identified as AI-generated.",[10,701,702,703,708],{},"MiniMax H3 uses its own ",[14,704,707],{"href":705,"rel":706},"https:\u002F\u002Fhuggingface.co\u002FEliovpAI\u002FMiniMax-H3-FP8-Paiton-RDNA4\u002Fblob\u002Fmain\u002FLICENSE",[79],"Community License",".\nIts standard grant excludes the EU, UK, South Korea and US, including use and display of outputs, and requires written\nauthorization for organizations above US$20 million in annual revenue. Check the full terms and any separate\npermission applicable to your use. A downloadable runtime is not unrestricted permission to use the model or its outputs.",[157,710,712],{"id":711},"guides-and-downloads","Guides and downloads",[123,714,715,721,727,734,740,747],{},[126,716,717],{},[14,718,720],{"href":77,"rel":719},[79],"LTX-2.5 setup and limits",[126,722,723],{},[14,724,726],{"href":492,"rel":725},[79],"LTX-2.5 FP8 weights",[126,728,729],{},[14,730,733],{"href":731,"rel":732},"https:\u002F\u002Fgithub.com\u002FEliovp-BV\u002Fpaiton-vllm-plugin\u002Freleases\u002Ftag\u002Fltx-2.5-rdna4-v0.1.0",[79],"LTX-2.5 v0.1.0 release",[126,735,736],{},[14,737,739],{"href":98,"rel":738},[79],"MiniMax H3 v1.2 release and launch instructions",[126,741,742],{},[14,743,746],{"href":744,"rel":745},"https:\u002F\u002Fhuggingface.co\u002FEliovpAI\u002FMiniMax-H3-FP8-Paiton-RDNA4",[79],"MiniMax H3 FP8 weights",[126,748,749],{},[14,750,752],{"href":751},"\u002Fproducts\u002Fpaiton#community","Paiton framework and community packages",[754,755,756],"style",{},"html pre.shiki code .sScJk, html code.shiki .sScJk{--shiki-default:#6F42C1;--shiki-dark:#B392F0}html pre.shiki code .sZZnC, html code.shiki .sZZnC{--shiki-default:#032F62;--shiki-dark:#9ECBFF}html pre.shiki code .sj4cs, html code.shiki .sj4cs{--shiki-default:#005CC5;--shiki-dark:#79B8FF}html pre.shiki code .szBVR, html code.shiki .szBVR{--shiki-default:#D73A49;--shiki-dark:#F97583}html pre.shiki code .sVt8B, html code.shiki .sVt8B{--shiki-default:#24292E;--shiki-dark:#E1E4E8}html pre.shiki code .sJ8bj, html code.shiki .sJ8bj{--shiki-default:#6A737D;--shiki-dark:#6A737D}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"title":502,"searchDepth":530,"depth":530,"links":758},[759,760,761,762,763,764],{"id":159,"depth":530,"text":160},{"id":180,"depth":530,"text":181},{"id":304,"depth":530,"text":305},{"id":473,"depth":530,"text":474},{"id":690,"depth":530,"text":155},{"id":711,"depth":530,"text":712},[17,766,767,768,80,769],"AMD Radeon","Local AI","Video Generation","MiniMax H3","2026-11-16T22:30:00Z","Paiton runs LTX-2.5 and MiniMax H3 on one R9700. A 5-second LTX clip with audio in 18.1 seconds, plus reference-image video, benchmarks and setup.","md",{},"https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-video-ltx-2-5-minimax-h3-radeon-r9700","\u002Fblog\u002Fpaiton-video-ltx-2-5-minimax-h3-radeon-r9700",{"title":5,"description":771},"paiton-video-ltx-2-5-minimax-h3-radeon-r9700","blog\u002Fpaiton-video-ltx-2-5-minimax-h3-radeon-r9700",null,{"name":781,"description":782,"src":174,"poster":173,"duration":783,"uploadDate":770},"LTX-2.5 street drummer with generated audio","AI-generated street drummer and crowd at dusk. A 1280 x 704 example made on one Radeon AI PRO R9700 with Paiton, including the model's audio.","PT5S","0yUBYpobEaClyc93T0D9UrNt9PVUOrnBifZ3B-_uGoY",[786,788,798,810,819,831,841,850,865,874,883,891,899,912,922,931,939,953,962,974,1005,1016,1038,1056,1075,1093,1110,1127,1138,1154,1169,1181,1190,1198,1213,1225,1236,1247,1257,1270,1280,1293,1304,1314,1325,1334,1346,1357,1366],{"path":775,"title":5,"description":771,"date":770,"slug":777,"image":173,"originalUrl":774,"categories":787},[17,766,767,768,80,769],{"path":36,"title":789,"description":790,"date":791,"slug":792,"image":793,"originalUrl":794,"categories":795},"MiniMax-H3-Image on a single R9700","MiniMax-H3-Image on one Radeon AI PRO R9700: 2048 x 2048 in 6.9 seconds, 4096 x 4096 in 33.4 seconds. FP8 weights, image editing and setup.","2026-11-12T10:00:00Z","paiton-minimax-h3-image-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-minimax-h3-image\u002F01-hero-concert.jpg","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-minimax-h3-image-radeon-ai-pro-r9700",[17,766,767,796,797],"Image Generation","MiniMax",{"path":799,"title":800,"description":801,"date":802,"slug":803,"image":804,"originalUrl":805,"categories":806},"\u002Fblog\u002Fpaiton-kimi-k31-pro-amd-instinct-mi355x","Kimi-K3.1-Pro on 6 MI355X GPUs at 800 W with Paiton","Kimi-K3.1-Pro on six MI355X GPUs at 800 W each. Paiton throughput, power measurements, a separate 1M-context test and a four-GPU option.","2026-11-11T12:00:00Z","paiton-kimi-k31-pro-amd-instinct-mi355x","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-kimi31\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-kimi-k31-pro-amd-instinct-mi355x",[17,807,808,809],"AMD Instinct","Inference","Kimi",{"path":41,"title":811,"description":812,"date":813,"slug":814,"image":815,"originalUrl":816,"categories":817},"The Radeon AI PRO R9700 does 3D too: Idol in 16.4 seconds","Idol turns an image into a 3D Gaussian splat in 16.44 seconds on one Radeon AI PRO R9700 with Paiton, versus 39.22 seconds for stock on the same GPU.","2026-11-06T09:00:00Z","paiton-idol-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-idol\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-idol-radeon-ai-pro-r9700",[17,766,767,818],"3D Generation",{"path":820,"title":821,"description":822,"date":823,"slug":824,"image":825,"originalUrl":826,"categories":827},"\u002Fblog\u002Fpaiton-deepseek-v41-flash-amd-instinct-mi350p","Deepseek V4.1 Flash on a single MI350P","DeepSeek V4.1 Flash on one 144 GB MI350P: 77.9 tokens\u002Fs for one request and 1,107.6 in a 512-request batch, with Paiton and CPU expert offload.","2026-11-03T12:00:00Z","paiton-deepseek-v41-flash-amd-instinct-mi350p","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-deepseek-v41\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-deepseek-v41-flash-amd-instinct-mi350p",[17,807,828,829,830],"AI Inference","DeepSeek","Quantization",{"path":832,"title":833,"description":834,"date":835,"slug":836,"image":837,"originalUrl":838,"categories":839},"\u002Fblog\u002Fpaiton-qwen38-flash-next-amd-instinct-mi350p","Qwen3.8 Flash Next on MI350P: twice the speed | Paiton","Qwen3.8 Flash Next on one AMD MI350P: 2.24× the tested vLLM setup's single-request decode speed and about 2.2 million shared KV-cache tokens with Paiton.","2026-10-30T12:00:00Z","paiton-qwen38-flash-next-amd-instinct-mi350p","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-mi350p\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-flash-next-amd-instinct-mi350p",[17,807,828,840,830],"Qwen",{"path":842,"title":843,"description":844,"date":845,"slug":846,"image":847,"originalUrl":848,"categories":849},"\u002Fblog\u002Fpaiton-qwen38-3bit-accuracy-prefix-caching-mtp","Qwen3.8 Flash Next: 3-bit near 16-bit accuracy, with MTP prefix caching","Paiton's new Qwen3.8 3-bit weights approach 16-bit task scores on two R9700 cards. Native MTP prefix caching resumes a 200K prompt in 0.68 seconds. Results, limits and setup.","2026-10-24T18:45:00Z","paiton-qwen38-3bit-accuracy-prefix-caching-mtp","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-qwen38-3bit-mtp\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-3bit-accuracy-prefix-caching-mtp",[17,766,767,840,830],{"path":851,"title":852,"description":853,"date":854,"slug":855,"image":856,"originalUrl":857,"categories":858},"\u002Fblog\u002Fnvidia-dsx-pod-vera-rubin-modular-data-centers","Nvidia confirms the modular DC direction","NVIDIA DSX POD brings Vera Rubin AI factories to modular data centers. ElioVP explains site readiness, cooling and the thinking behind ModFlex.","2026-10-21T13:00:00Z","nvidia-dsx-pod-vera-rubin-modular-data-centers","\u002Fasset\u002Fimages\u002Fblog\u002Fnvidia-dsx-pod-modflex\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fnvidia-dsx-pod-vera-rubin-modular-data-centers",[859,860,861,862,863,864],"Modular DC","AI Infrastructure","ModFlex","NVIDIA","Vera Rubin","Liquid Cooling",{"path":866,"title":867,"description":868,"date":869,"slug":870,"image":871,"originalUrl":872,"categories":873},"\u002Fblog\u002Fpaiton-qwen38-flash-next-4bit-radeon-ai-pro-r9700","Qwen3.8 Flash Next in 4-bit on 2 × Radeon R9700: tested against 16-bit","Paiton runs Qwen3.8 Flash Next in 4-bit on two R9700 cards: 96.8% GSM8K, 96.3% HumanEval, 189.8 tokens\u002Fs and an optional RAM prefix cache. Accuracy, speed and setup.","2026-10-12T15:00:00Z","paiton-qwen38-flash-next-4bit-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-qwen38-flash-next-4bit\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-flash-next-4bit-radeon-ai-pro-r9700",[17,766,767,840,830],{"path":875,"title":876,"description":877,"date":878,"slug":879,"image":880,"originalUrl":881,"categories":882},"\u002Fblog\u002Fpaiton-qwen38-flash-next-radeon-ai-pro-r9700","Qwen3.8 Flash Next on 2 × Radeon AI PRO R9700: 216 tokens\u002Fs, 200K context","Paiton runs Qwen3.8 Flash Next with 3-bit experts on two R9700 cards: 216.3 tokens\u002Fs, 565.1 aggregate tokens\u002Fs and a separate 200K mode. Benchmarks, quality and setup.","2026-10-10T10:30:00Z","paiton-qwen38-flash-next-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-qwen38-flash-next\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-flash-next-radeon-ai-pro-r9700",[17,766,767,840,830],{"path":21,"title":884,"description":885,"date":886,"slug":887,"image":888,"originalUrl":889,"categories":890},"Qwen3.8 27B on 1 × Radeon AI PRO R9700: 3-bit weights, 20% faster decode","Qwen3.8 27B on one R9700: 19.9% faster weighted decode, a new 4-bit cache, 200K context and optional vision. Benchmarks and current Paiton setup.","2026-09-27T09:00:00Z","paiton-qwen38-w3a4-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-qwen38-w3a4\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-w3a4-radeon-ai-pro-r9700",[17,766,767,840,830],{"path":31,"title":892,"description":893,"date":894,"slug":895,"image":896,"originalUrl":897,"categories":898},"Qwen-Image 2.1 on 1 × Radeon AI PRO R9700: 2048×2048 images in 103 seconds","Generate 2048×2048 Qwen-Image 2.1 images locally with 1 × Radeon AI PRO R9700. The released Paiton v1.0.2 container measured 103.29 seconds per warm request, through PNG delivery.","2026-09-23T09:00:00Z","paiton-qwen-image-21-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-qwen-image-21\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen-image-21-radeon-ai-pro-r9700",[17,766,767,796,840],{"path":900,"title":901,"description":902,"date":903,"slug":904,"image":905,"originalUrl":906,"categories":907},"\u002Fblog\u002Fpaiton-qwen38-mxfp4-dflash2-r9700","Qwen3.8: 400.7 tok\u002Fs on R9700 | Paiton","Qwen3.8 on one R9700: 400.7 aggregate tok\u002Fs with ROCm 10 and vLLM 0.29, plus public 200K\u002F220K chat profiles. Benchmarks, limits and launch commands.","2026-09-16T07:30:00Z","paiton-qwen38-mxfp4-dflash2-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-qwen38-mxfp4\u002Fupdate-2026-09-19\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-mxfp4-dflash2-r9700",[17,766,908,909,910,911],"vLLM","Qwen3.8","Inference Optimization","DFlash2",{"path":913,"title":914,"description":915,"date":916,"slug":917,"image":918,"originalUrl":919,"categories":920},"\u002Fblog\u002Fpaiton-qwen38-neo-gguf-vllm-r9700","Qwen3.8 GGUF in vLLM: Faster Responses on One Radeon","Run the original NEO CODER MAX GGUF in vLLM with Paiton on an R9700. Explore measured latency gains, image input and local deployment.","2026-09-14T07:30:00Z","paiton-qwen38-neo-gguf-vllm-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-neo-gguf\u002F00-hero-neo-gguf-r9700.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-neo-gguf-vllm-r9700",[17,766,767,921,908],"GGUF",{"path":468,"title":923,"description":924,"date":925,"slug":926,"image":927,"originalUrl":928,"categories":929},"MiniMax H3 on Radeon: 15-Second Video With Native Sound","Paiton generates a 15-second MiniMax H3 video with stereo audio on one Radeon AI PRO R9700 in 5m 33s, with 16.7% lower latency than matched stock.","2026-09-09T07:30:00Z","paiton-minimax-h3-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-minimax-h3\u002F00-featured-minimax-h3-r9700.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-minimax-h3-radeon-ai-pro-r9700",[17,766,767,768,769,930],"ComfyUI",{"path":26,"title":932,"description":933,"date":934,"slug":935,"image":936,"originalUrl":26,"categories":937},"Local FLUX.2 klein on Radeon AI PRO R9700: Faster Image Generation with Less VRAM","Paiton generates 1024×1024 FLUX.2 klein images in 1.054 seconds on an R9700: 16.2% lower warm latency and 33.4% lower peak Torch allocation.","2026-09-07T09:00:00","paiton-flux2-klein-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-flux2-klein\u002Ffox-paiton.webp",[17,766,767,796,938,930],"FLUX",{"path":940,"title":941,"description":942,"date":943,"slug":944,"image":945,"originalUrl":946,"categories":947},"\u002Fblog\u002Fpaiton-ornith15-radeon-ai-pro-r9700","Ornith 1.5 at 44.6 tok\u002Fs on One Radeon AI PRO R9700","Paiton serves Ornith 1.5 35B A3B at 44.63 output tok\u002Fs on one Radeon AI PRO R9700, 27% faster than tuned stock vLLM, with 21.3% lower modeled cost.","2026-09-05T09:00:00","paiton-ornith15-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-ornith15\u002F00-featured-ornith15-r9700.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-ornith15-radeon-ai-pro-r9700",[17,948,766,828,949,950,910,951,908,952],"Artificial Intelligence","GPU Performance","Inference Latency","Large Language Models","Cost Efficiency",{"path":954,"title":955,"description":956,"date":957,"slug":958,"image":959,"originalUrl":960,"categories":961},"\u002Fblog\u002Fpaiton-qwen38-radeon-ai-pro-r9700","Paiton: 21% More Qwen3.8 Throughput on Radeon AI PRO R9700","Paiton serves AMD’s Qwen3.8 27B at 39.77 output tokens\u002Fs on one Radeon AI PRO R9700, delivering 21% more throughput and 17.4% lower modeled cost.","2026-09-04T09:00:00","paiton-qwen38-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-r9700\u002F00-featured-paiton-r9700.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-radeon-ai-pro-r9700",[17,948,766,828,949,950,910,951,908,952],{"path":963,"title":964,"description":965,"date":966,"slug":967,"image":968,"originalUrl":779,"categories":969},"\u002Fblog\u002Fai-data-center-power-requirements-gpu-per-megawatt","AI Data Center Power Requirements: The GPU\u002FMW Illusion","Why do AI data-center proposals quote different GPU capacities? See how PUE, peak loads, storage, networking and cooling determine deployable compute.","2026-07-27T23:52:00","ai-data-center-power-requirements-gpu-per-megawatt","\u002Fasset\u002Fimages\u002Fblog\u002Fai-data-center-power-requirements-gpu-per-megawatt\u002Fgpu-per-megawatt-illusion.webp",[970,860,971,861,972,973],"All","Data Centers","HPC","AMD Helios",{"path":975,"title":976,"description":977,"date":978,"slug":979,"image":980,"originalUrl":981,"categories":982},"\u002Fblog\u002Fpaiton-returns-to-its-diffusion-roots-optimizing-wan2-2-t2v-a14b-on-amd-mi355x","Wan2.2 Video Generation: Paiton on AMD MI355X","Compare Wan2.2-T2V-A14B video generation on AMD MI355X with Paiton and NVIDIA B200 using Diffusers, and explore our diffusion optimization approach.","2026-06-10T14:04:04","paiton-returns-to-its-diffusion-roots-optimizing-wan2-2-t2v-a14b-on-amd-mi355x","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fpaitonwan2.webp","https:\u002F\u002Feliovp.com\u002Fpaiton-returns-to-its-diffusion-roots-optimizing-wan2-2-t2v-a14b-on-amd-mi355x\u002F",[970,948,17,983,984,985,986,987,988,989,990,991,992,993,808,994,995,996,997,998,17,999,1000,1001,1002,1003,1004],"14B","AMD","B200","Benchmarks","Blackwell","Compute","Diffusion","Eliovp","Generative AI","GPU","Hardware","Instinct","MI355x","NVidia","On-Premise","Optimization","Sovereign AI","T2V","Text-to-Video","Tuning","Video-Generation","Wan2.2",{"path":1006,"title":1007,"description":1008,"date":1009,"slug":1010,"image":1011,"originalUrl":1012,"categories":1013},"\u002Fblog\u002Ffrom-the-attic-to-the-front-page-eliovp-recognized-as-a-pioneer-in-chip-optimization-data-center-infrastructure","ElioVP in De Tijd: Chip Optimization and Data Centers","Read about De Tijd's coverage of ElioVP, from its origins in chip optimization to its work on modular data centers and high-density cooling.","2026-02-10T20:48:12","from-the-attic-to-the-front-page-eliovp-recognized-as-a-pioneer-in-chip-optimization-data-center-infrastructure","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fphysicalnewspaper.webp","https:\u002F\u002Feliovp.com\u002Ffrom-the-attic-to-the-front-page-eliovp-recognized-as-a-pioneer-in-chip-optimization-data-center-infrastructure\u002F",[970,948,859,1014,984,1015,859,996],"Uncategorized","De Tijd",{"path":1017,"title":1018,"description":1019,"date":1020,"slug":1021,"image":1022,"originalUrl":1023,"categories":1024},"\u002Fblog\u002Fprivacy-is-geen-it-probleem-meer-het-is-een-strategische-prioriteit","AI Privacy: A Strategic Priority for Benelux Businesses","Explore generative AI privacy risks, trust, data retention and governance, and why Benelux businesses need a strategic approach to secure AI.","2026-01-29T13:51:11","privacy-is-geen-it-probleem-meer-het-is-een-strategische-prioriteit","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fheaderimage.webp","https:\u002F\u002Feliovp.com\u002Fprivacy-is-geen-it-probleem-meer-het-is-een-strategische-prioriteit\u002F",[970,948,1025,1014,1026,1027,1028,1029,1030,1031,1032,1033,990,1034,991,1035,1036,1037],"Trending","AI Act","Anthropomorphism","AVG","Benelux","ChatGPT","Cybersecurity","Data Governance","Data Security","GDPR","Microsoft Copilot","Privacy","Shadow AI",{"path":1039,"title":1040,"description":1041,"date":1042,"slug":1043,"image":1044,"originalUrl":1045,"categories":1046},"\u002Fblog\u002Fitsme-bij-ons-is-het-its-not-me-en-dit-is-waarom","Why We Do Not Use itsme: Privacy and Data Sovereignty","Why ElioVP does not use itsme: our assessment of identity metadata, cloud dependence, data sovereignty and authentication risks.","2025-11-27T09:32:14","itsme-bij-ons-is-het-its-not-me-en-dit-is-waarom","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Ffrontimage.webp","https:\u002F\u002Feliovp.com\u002Fitsme-bij-ons-is-het-its-not-me-en-dit-is-waarom\u002F",[970,1047,1048,1049,1031,1050,1051,1052,1034,1053,1054,1055,1036],"AWS","Belgian Mobile ID","Cloud Act","Data Sovereignty","Digital Identity","eIDAS","itsme","Liberty Global","MyGov.be",{"path":1057,"title":1058,"description":1059,"date":1060,"slug":1061,"image":1062,"originalUrl":1063,"categories":1064},"\u002Fblog\u002Ffield-report-the-reality-of-building-agentic-ai-in-2025","Field Report. The Reality of Building Agentic AI in 2025","Lessons from building on-premise AI agents in 2025 cover workflow design, observability, model training, hallucinations and GPU memory limits.","2025-11-25T14:03:39","field-report-the-reality-of-building-agentic-ai-in-2025","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Ffieldreport.webp","https:\u002F\u002Feliovp.com\u002Ffield-report-the-reality-of-building-agentic-ai-in-2025\u002F",[970,948,1065,1025,1066,1067,1068,1069,1070,1071,1072,1073,999,1074],"Solutions","Agentic AI","AI Engineering","AI Strategy","Autonomous Agents","Enterprise AI","Local LLM","Model Fine-Tuning","On-Premise AI","VRAM Optimization",{"path":1076,"title":1077,"description":1078,"date":1079,"slug":1080,"image":1081,"originalUrl":1082,"categories":1083},"\u002Fblog\u002Fthe-synthetic-unicorn-bubble","The Synthetic Unicorn Bubble","An analysis of AI neocloud investment risks, examining circular financing, infrastructure claims, contract terms and due diligence.","2025-11-24T19:22:28","the-synthetic-unicorn-bubble","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fsyntheticunicorn.webp","https:\u002F\u002Feliovp.com\u002Fthe-synthetic-unicorn-bubble\u002F",[970,948,1025,860,1084,1085,1086,1087,1088,1089,1090,1091,1092],"AI Neocloud","Circular Financing","GPU Cloud","Investment Risks","Startup Valuation","Synthetic Bubble","Tech Analysis","Vaporware","Venture Capital",{"path":1094,"title":1095,"description":1096,"date":1097,"slug":1098,"image":1099,"originalUrl":1100,"categories":1101},"\u002Fblog\u002Fbuilding-the-engine-for-the-ai-race-the-4-month-path-to-nvidia-gb300-nvl72-power","NVIDIA GB300 NVL72: A Four-Month Modular Data Center Plan","Explore a modular data center design for NVIDIA GB300 NVL72, covering redundant power, hybrid cooling and a four-month deployment plan.","2025-11-20T14:10:19","building-the-engine-for-the-ai-race-the-4-month-path-to-nvidia-gb300-nvl72-power","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fsuperpodmodflexfrontimage.webp","https:\u002F\u002Feliovp.com\u002Fbuilding-the-engine-for-the-ai-race-the-4-month-path-to-nvidia-gb300-nvl72-power\u002F",[970,859,1014,1102,860,1103,1104,864,1105,1106,1107,1108,1109],"150kW Rack","DLC","High Density","Modular Data Center","NVIDIA Blackwell Ultra","NVIDIA GB300","NVL72","Rapid Deployment",{"path":1111,"title":1112,"description":1113,"date":1114,"slug":1115,"image":1116,"originalUrl":1117,"categories":1118},"\u002Fblog\u002Fwhy-cuda-translation-wont-unlock-amds-real-potential","Why “CUDA” Translation Won’t Unlock AMD’s Real Potential","Why CUDA compatibility is not the same as AMD performance: explore ROCm, HIP, kernel tuning and the case for hardware-specific optimization.","2025-11-12T14:48:37","why-cuda-translation-wont-unlock-amds-real-potential","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fchatgpt-image-nov-11-2025-09_16_10-pm-1.webp","https:\u002F\u002Feliovp.com\u002Fwhy-cuda-translation-wont-unlock-amds-real-potential\u002F",[970,948,17,1014,1119,948,1120,1121,1122,1123,1124,1125,17,1126],"AMD MI300X","CUDA Translation","FP8","GPU Optimization","High Performance Computing","HIP","Kernel Tuning","ROCm",{"path":1128,"title":1129,"description":1130,"date":1131,"slug":1132,"image":1133,"originalUrl":1134,"categories":1135},"\u002Fblog\u002Fpaiton-the-simplest-way-to-supercharge-ai-inference","Paiton: The Simplest Way to Supercharge AI Inference","Learn how Paiton integrates with existing inference stacks, with AMD MI300X benchmark results and performance-per-dollar comparisons.","2025-11-11T10:31:22","paiton-the-simplest-way-to-supercharge-ai-inference","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fpaiton-powaaah.webp","https:\u002F\u002Feliovp.com\u002Fpaiton-the-simplest-way-to-supercharge-ai-inference\u002F",[970,948,17,828,807,1119,952,1136,910,1125,17,1137,908],"High Throughput","SGLang",{"path":1139,"title":1140,"description":1141,"date":1142,"slug":1143,"image":1144,"originalUrl":1145,"categories":1146},"\u002Fblog\u002Fstop-overpaying-paiton-mi300x-moe-beats-h200-b200-on-1m-tokens","Paiton MoE Benchmarks: MI300X vs H200 and B200","Compare Qwen3-30B-A3B MoE inference with Paiton on MI300X against H200 and B200, including throughput and cost per million tokens.","2025-09-26T13:36:18","stop-overpaying-paiton-mi300x-moe-beats-h200-b200-on-1m-tokens","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fhulkvshulkpaitonwins.webp","https:\u002F\u002Feliovp.com\u002Fstop-overpaying-paiton-mi300x-moe-beats-h200-b200-on-1m-tokens\u002F",[970,948,17,1147,1119,1148,910,1149,1150,1151,1152,17,1153],"AI Benchmarks","Cost per Token","Mixture of Experts","MoE","NVIDIA B200","NVIDIA H200","Qwen3",{"path":1155,"title":1156,"description":1157,"date":1158,"slug":1159,"image":1160,"originalUrl":1161,"categories":1162},"\u002Fblog\u002Fagentic-ai-but-make-it-local-from-inbox-to-insight-to-action-en","Local Agentic AI: From Inbox to Action","Local-first AI agents turn email, documents and images into tickets, reports and actions, using models tailored to your data and systems.","2025-09-16T13:09:00","agentic-ai-but-make-it-local-from-inbox-to-insight-to-action-en","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Ffrontfotoblog.webp","https:\u002F\u002Feliovp.com\u002Fagentic-ai-but-make-it-local-from-inbox-to-insight-to-action-en\u002F",[970,948,1065,1014,1066,1163,1164,1165,1166,1071,1073,999,1167,1168],"Damage Detection","Document Processing","Email Automation","Invoice Extraction","Ticket Automation","Workflow Automation",{"path":1170,"title":1171,"description":1172,"date":1173,"slug":1174,"image":1175,"originalUrl":1176,"categories":1177},"\u002Fblog\u002Fmi300x-fp8-data-parallel-benchmarks-8-64-gpus-h200-left-behind-b200-within-reach","MI300X FP8 Benchmarks: GPU Partitioning with Paiton","Explore Llama 3.1 8B FP8 benchmarks on partitioned MI300X GPUs with Paiton, comparing throughput and latency against NVIDIA H200 and B200.","2025-07-31T13:32:57","mi300x-fp8-data-parallel-benchmarks-8-64-gpus-h200-left-behind-b200-within-reach","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fimage-2-1.webp","https:\u002F\u002Feliovp.com\u002Fmi300x-fp8-data%e2%80%91parallel-benchmarks-8-64-gpus-h200-left-behind-b200-within-reach\u002F",[970,948,17,1014,1178,984,985,1179,1180,995,996,17,908],"AI","H200","MI300X",{"path":1182,"title":1183,"description":1184,"date":1185,"slug":1186,"image":1187,"originalUrl":1188,"categories":1189},"\u002Fblog\u002Fapplicable-ai-for-businesses","Applicable AI for Businesses","Explore ElioVP's approach to local AI for business workflows, including custom model training and automated damage detection for logistics.","2025-07-09T21:35:30","applicable-ai-for-businesses","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fscherm_afbeelding-2025-07-09-om-23.30.17.webp","https:\u002F\u002Feliovp.com\u002Fapplicable-ai-for-businesses\u002F",[970,948,1065,1066,1163,1164,1165,1166,1071,1073,999,1167,1168],{"path":1191,"title":1192,"description":1193,"date":1194,"slug":1195,"image":502,"originalUrl":1196,"categories":1197},"\u002Fblog\u002Fintroducing-paitons-free-evaluation-models","Introducing Paiton’s Free Evaluation Models","Test Paiton with free evaluation models for AMD GPUs. Compare text, vision and image generation performance using your own workloads.","2025-07-07T11:26:13","introducing-paitons-free-evaluation-models","https:\u002F\u002Feliovp.com\u002Fintroducing-paitons-free-evaluation-models\u002F",[970,948,17],{"path":1199,"title":1200,"description":1201,"date":1202,"slug":1203,"image":1204,"originalUrl":1205,"categories":1206},"\u002Fblog\u002Fpaiton-dramatically-faster-startup-and-performance-for-llama-3-1-405b","Llama 3.1 405B: Faster Startup with Paiton on MI300X","See Paiton benchmarks for Llama 3.1 405B on eight AMD MI300X GPUs, covering model startup, tensor parallelism, throughput and latency.","2025-06-12T20:15:23","paiton-dramatically-faster-startup-and-performance-for-llama-3-1-405b","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fservingscreenshot.webp","https:\u002F\u002Feliovp.com\u002Fpaiton-dramatically-faster-startup-and-performance-for-llama-3-1-405b\u002F",[970,948,17,1014,828,1119,1207,1121,1208,1209,1210,17,1211,1212],"Cold Start","Graph Compilation","Llama 3.1 405B","LLM Optimization","Startup Latency","Tensor Parallelism",{"path":1214,"title":1215,"description":1216,"date":1217,"slug":1218,"image":1219,"originalUrl":1220,"categories":1221},"\u002Fblog\u002Fpaiton-fp8-beats-nvidias-h200-on-amds-mi300x","Paiton FP8 Beats NVIDIA’s H200 on AMD’s MI300X","Compare Paiton on AMD MI300X with NVIDIA H200 for Llama 3.1 70B FP8, including throughput, first-token delay and latency across batch sizes.","2025-06-08T19:12:40","paiton-fp8-beats-nvidias-h200-on-amds-mi300x","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fblognewfp8.webp","https:\u002F\u002Feliovp.com\u002Fpaiton-fp8-beats-nvidias-h200-on-amds-mi300x\u002F",[970,948,17,1014,1119,1222,1070,991,949,950,951,1209,1223,1224],"Cold Start Optimization","Model Serving","vLLM Optimization",{"path":1226,"title":1227,"description":1228,"date":1229,"slug":1230,"image":1231,"originalUrl":1232,"categories":1233},"\u002Fblog\u002Fmi300x-vs-h200-vs-rx-7900-xtx-vs-tenstorrent-n300s-with-vllm","MI300X vs H200 vs RX 7900 XTX vs Tenstorrent n300s with vLLM","Compare MI300X, H200, RX 7900 XTX and Tenstorrent n300s on Llama 3 8B with vLLM, including throughput, modeled token costs and hardware limits.","2025-05-09T14:03:58","mi300x-vs-h200-vs-rx-7900-xtx-vs-tenstorrent-n300s-with-vllm","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fcomparisontenstor.webp","https:\u002F\u002Feliovp.com\u002Fmi300x-vs-h200-vs-rx-7900-xtx-vs-tenstorrent-n300s-with-vllm\u002F",[970,948,17,1065,1014,984,1180,996,1234,1235],"RX7900XTX","tenstorrent",{"path":1237,"title":1238,"description":1239,"date":1240,"slug":1241,"image":1242,"originalUrl":1243,"categories":1244},"\u002Fblog\u002Fclusterpl-empowering-gpu-cluster-investors-with-real-world-financial-insights","ClusterP&L: Financial Modeling for GPU Clusters","Explore how ClusterP&L models GPU cluster costs, profitability and investment scenarios, with ROI metrics, risk simulations and exportable reports.","2025-05-03T10:52:22","clusterpl-empowering-gpu-cluster-investors-with-real-world-financial-insights","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fcomparisonscenarios.webp","https:\u002F\u002Feliovp.com\u002Fclusterpl-empowering-gpu-cluster-investors-with-real-world-financial-insights\u002F",[970,948,859,1065,985,1179,1245,996,1246],"MI325x","pnl calculator",{"path":1248,"title":1249,"description":1250,"date":1251,"slug":1252,"image":1253,"originalUrl":1254,"categories":1255},"\u002Fblog\u002Fcranking-out-faster-tokens-for-fewer-dollars-amd-mi300x-vs-nvidia-h200","AMD MI300X vs. NVIDIA H200: Qwen3-32B with Paiton","Compare Qwen3-32B benchmarks on Paiton-optimized AMD MI300X and NVIDIA H200, covering throughput, latency and hardware costs.","2025-05-02T21:10:30","cranking-out-faster-tokens-for-fewer-dollars-amd-mi300x-vs-nvidia-h200","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002F3ac59a73-2466-4422-b7e5-ef2e4a8ca58e.webp","https:\u002F\u002Feliovp.com\u002Fcranking-out-faster-tokens-for-fewer-dollars-amd-mi300x-vs-nvidia-h200\u002F",[970,948,17,1178,984,1179,1256,996,17,908],"MI300",{"path":1258,"title":1259,"description":1260,"date":1261,"slug":1262,"image":1263,"originalUrl":1264,"categories":1265},"\u002Fblog\u002Fpower-meets-precision-high-density-modular-data-center-for-nvidia-nvl-deployments-1-2-mw","Modular Data Centers for NVIDIA NVL: 1 to 2 MW","Explore modular data center designs for NVIDIA NVL systems, covering power capacity, liquid cooling, redundancy and deployment planning.","2025-05-02T14:09:59","power-meets-precision-high-density-modular-data-center-for-nvidia-nvl-deployments-1-2-mw","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Feliovp_critical-1mw-pod_rev-2_transparent.webp","https:\u002F\u002Feliovp.com\u002Fpower-meets-precision-high-density-modular-data-center-for-nvidia-nvl-deployments-1-2-mw\u002F",[970,859,1266,860,1104,972,864,1105,1267,1268,1108,1269],"1-2MW Data Center","NVIDIA Blackwell","NVIDIA NVL","Precision Cooling",{"path":1271,"title":1272,"description":1273,"date":1274,"slug":1275,"image":1276,"originalUrl":1277,"categories":1278},"\u002Fblog\u002Fexamining-ai-agents-in-the-medical-field-ai-that-speaks-dicom","Examining AI agents in the medical field: AI that speaks DICOM","Explore a local AI agent demo for DICOM workflows, from patient and study retrieval to a comparison of vision models using anonymized medical images.","2025-04-11T14:45:48","examining-ai-agents-in-the-medical-field-ai-that-speaks-dicom","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fhealthcareblog-1.webp","https:\u002F\u002Feliovp.com\u002Fexamining-ai-agents-in-the-medical-field-ai-that-speaks-dicom\u002F",[970,948,1065,1014,1178,984,1279],"Healthcare",{"path":1281,"title":1282,"description":1283,"date":1284,"slug":1285,"image":1286,"originalUrl":1287,"categories":1288},"\u002Fblog\u002Feliovp-bv-your-trusted-partner-for-supply-chain-resilience-amidst-new-u-s-tariffs","U.S. Tariffs and AI Supply Chain Resilience: April 2025","Read ElioVP's April 2025 perspective on U.S. tariffs and supply chain resilience for AI servers, HPC systems and modular data centers.","2025-04-04T10:01:27","eliovp-bv-your-trusted-partner-for-supply-chain-resilience-amidst-new-u-s-tariffs","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Ftariffsshipping.webp","https:\u002F\u002Feliovp.com\u002Feliovp-bv-your-trusted-partner-for-supply-chain-resilience-amidst-new-u-s-tariffs\u002F",[970,1025,1178,984,1289,1290,1291,1292],"import","Taiwan","Tariffs","Trump",{"path":1294,"title":1295,"description":1296,"date":1297,"slug":1298,"image":1299,"originalUrl":1300,"categories":1301},"\u002Fblog\u002Fwhy-ai-agents-are-the-future","Why AI Agents Are the Future","Explore AI agents for ERP, CRM, finance and customer support, with practical use cases and a path from workflow assessment to pilot and deployment.","2025-03-23T22:06:59","why-ai-agents-are-the-future","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Ferp2.jpeg","https:\u002F\u002Feliovp.com\u002Fwhy-ai-agents-are-the-future\u002F",[970,948,1065,1178,1302,1303],"AI Agents","ERP",{"path":1305,"title":1306,"description":1307,"date":1308,"slug":1309,"image":1310,"originalUrl":1311,"categories":1312},"\u002Fblog\u002Fthe-rise-of-open-source-ai-model-optimization","The Rise of Open-Source AI Model Optimization","Explore open-source AI optimization trends, from quantization and mixture-of-experts models to hardware-aware tuning, RAG and edge deployment.","2025-03-22T20:59:23","the-rise-of-open-source-ai-model-optimization","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Friseofopensource.jpeg","https:\u002F\u002Feliovp.com\u002Fthe-rise-of-open-source-ai-model-optimization\u002F",[970,948,1025,1313,984,992,996],"AI news",{"path":1315,"title":1316,"description":1317,"date":1318,"slug":1319,"image":1320,"originalUrl":1321,"categories":1322},"\u002Fblog\u002Fintroducing-our-benchmarking-tool-powered-by-dstack","Introducing Our Benchmarking Tool: Powered by dstack","Explore our dstack-powered tool for reproducible vLLM benchmarks, automated parameter sweeps and performance reports across local and cloud GPUs.","2025-03-20T14:21:59","introducing-our-benchmarking-tool-powered-by-dstack","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fbenchmarktool.jpeg","https:\u002F\u002Feliovp.com\u002Fintroducing-our-benchmarking-tool-powered-by-dstack\u002F",[970,948,17,1178,984,1323,1324,1180,17],"benchmark","LLM",{"path":1326,"title":1327,"description":1328,"date":1329,"slug":1330,"image":1331,"originalUrl":1332,"categories":1333},"\u002Fblog\u002Foptimizing-qwq-32b-by-qwen-amd-mi300x-vs-nvidia-h200","Optimizing QwQ-32B (by Qwen): AMD MI300X vs. NVIDIA H200","Compare QwQ-32B throughput and latency on AMD MI300X with Paiton and NVIDIA H200, from small batches to higher concurrency.","2025-03-19T21:41:44","optimizing-qwq-32b-by-qwen-amd-mi300x-vs-nvidia-h200","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fpaiton4.jpeg","https:\u002F\u002Feliovp.com\u002Foptimizing-qwq-32b-by-qwen-amd-mi300x-vs-nvidia-h200\u002F",[970,948,17],{"path":1335,"title":1336,"description":1337,"date":1338,"slug":1339,"image":1340,"originalUrl":1341,"categories":1342},"\u002Fblog\u002Feliovp-featured-on-amd-tech-talk-podcast","Eliovp Featured on AMD “Tech Talk” Podcast","Listen to Elio Van Puyvelde and Jim Greene on AMD's Tech Talk podcast, discussing ElioVP's origins and its AI hardware and software services.","2025-03-19T07:53:39","eliovp-featured-on-amd-tech-talk-podcast","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Ftechtalkjimgreene.jpeg","https:\u002F\u002Feliovp.com\u002Feliovp-featured-on-amd-tech-talk-podcast\u002F",[970,984,1343,1344,1345],"Jim Greene","Podcast","Tech Talk",{"path":1347,"title":1348,"description":1349,"date":1350,"slug":1351,"image":1352,"originalUrl":1353,"categories":1354},"\u002Fblog\u002Ffurther-optimizing-amd-powered-inference-with-paiton","Further Optimizing AMD-Powered Inference with Paiton","Explore Paiton's DeepSeek R1 Distill Llama 8B benchmarks on AMD MI300X, focusing on throughput and first-token latency at smaller batch sizes.","2025-03-13T06:18:30","further-optimizing-amd-powered-inference-with-paiton","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fpaitonpost3.webp","https:\u002F\u002Feliovp.com\u002Ffurther-optimizing-amd-powered-inference-with-paiton\u002F",[970,948,17,984,1355,1356,1179,1180,1245,17,908],"Deepseek","H100",{"path":1358,"title":1359,"description":1360,"date":1361,"slug":1362,"image":1363,"originalUrl":1364,"categories":1365},"\u002Fblog\u002Fa-first-look-at-paiton-in-action-deepseek-r1-distill-llama-3-1-8b","Paiton Benchmarks: DeepSeek R1 Distill Llama 3.1 8B","Compare stock and Paiton-optimized DeepSeek R1 Distill Llama 3.1 8B on AMD MI300X, with throughput and latency benchmarks across batch sizes.","2025-01-31T09:11:02","a-first-look-at-paiton-in-action-deepseek-r1-distill-llama-3-1-8b","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fpaitonpost2.webp","https:\u002F\u002Feliovp.com\u002Fa-first-look-at-paiton-in-action-deepseek-r1-distill-llama-3-1-8b\u002F",[970,948,17,984,1355,1356,1179,1180,1245,17,908],{"path":1367,"title":1368,"description":1369,"date":1370,"slug":1371,"image":1372,"originalUrl":1373,"categories":1374},"\u002Fblog\u002Fai-model-optimization-with-paiton","AI Model Optimization with Paiton","Learn how Paiton uses model compilation, custom kernels and kernel fusion to optimize AI inference on AMD GPUs.","2025-01-30T19:53:25","ai-model-optimization-with-paiton","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fpaitonpost1.webp","https:\u002F\u002Feliovp.com\u002Fai-model-optimization-with-paiton\u002F",[970,948,17,984,1356,1179,1180,1245,17,908],1794869699013]