https://nodegrove.io/ What Nodegrove measured on rented GPUs, and the data sheets, tools and guides whose numbers changed, newest first. en Fri, 23 Oct 2026 00:00:00 GMT https://nodegrove.io/benchmarks tag:nodegrove.io,2026-10-23:/benchmarks Fri, 23 Oct 2026 00:00:00 GMT Updated 23 October 2026. What the jobs we ran took and cost on each card, our memory maths against the measured peaks, and how we measure. https://nodegrove.io/guides/dgx-spark-vs-strix-halo-vs-mac-studio tag:nodegrove.io,2026-10-23:/guides/dgx-spark-vs-strix-halo-vs-mac-studio Fri, 23 Oct 2026 00:00:00 GMT Updated 23 October 2026. The three big-memory boxes for local LLMs: memory the GPU can use, bandwidth, software, and speed on the same models. https://nodegrove.io/guides/local-ai-server tag:nodegrove.io,2026-10-23:/guides/local-ai-server Fri, 23 Oct 2026 00:00:00 GMT Updated 23 October 2026. The three build tiers, the running costs nobody lists, and a break-even calculator with your own numbers. https://nodegrove.io/guides/openclaw-vram-requirements tag:nodegrove.io,2026-10-23:/guides/openclaw-vram-requirements Fri, 23 Oct 2026 00:00:00 GMT Updated 23 October 2026. The context OpenClaw asks for, what it costs in memory, and the best tool-calling model for 12, 16 and 24 GB. https://nodegrove.io/guides/personal-ai-computer tag:nodegrove.io,2026-10-23:/guides/personal-ai-computer Fri, 23 Oct 2026 00:00:00 GMT Updated 23 October 2026. Ghost Core, DGX Spark, Strix Halo, Mac Studio and RTX PCs: the memory each GPU can use, its speed, and the models each one runs, agents included. https://nodegrove.io/guides/private-ai-server tag:nodegrove.io,2026-10-23:/guides/private-ai-server Fri, 23 Oct 2026 00:00:00 GMT Updated 23 October 2026. What "private" means at each layer, and the three ways to get one, honestly compared. https://nodegrove.io/guides/self-hosted-llm tag:nodegrove.io,2026-10-23:/guides/self-hosted-llm Fri, 23 Oct 2026 00:00:00 GMT Updated 23 October 2026. How VRAM is really used, the runtimes that matter, quantisation without the mysticism, and a calculator. https://nodegrove.io/guides/what-fits-on-24-48-96-gb tag:nodegrove.io,2026-10-23:/guides/what-fits-on-24-48-96-gb Fri, 23 Oct 2026 00:00:00 GMT Updated 23 October 2026. Estimated VRAM for popular open models at Q4 and Q8, 8k and 32k context, with the smallest card that fits. https://nodegrove.io/gpus tag:nodegrove.io,2026-10-22:/gpus Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. Every card against the same open models: the pick at each memory size, what fits at Q4 and how fast each answers. https://nodegrove.io/where-to-run tag:nodegrove.io,2026-10-22:/where-to-run Thu, 22 Oct 2026 00:00:00 GMT Published 22 October 2026. Your own GPU, Comfy Cloud, a cloud GPU by the hour or a per-clip API: what each costs, and when each one wins. https://nodegrove.io/tools/can-i-run-it tag:nodegrove.io,2026-10-22:/tools/can-i-run-it Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. Your card, any model: fits or not, how fast, and what to change when it does not. https://nodegrove.io/tools/comfyui-vram-checker tag:nodegrove.io,2026-10-22:/tools/comfyui-vram-checker Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. Any open image or video model on your card: which file to download, stage by stage. https://nodegrove.io/tools/comfyui-workflow-checker tag:nodegrove.io,2026-10-22:/tools/comfyui-workflow-checker Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. Drop a workflow: every model file it loads, the VRAM it needs, what to swap and the custom nodes. https://nodegrove.io/tools/llm-speed-estimator tag:nodegrove.io,2026-10-22:/tools/llm-speed-estimator Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. How fast a model runs on a given GPU, from bandwidth and weight size. https://nodegrove.io/tools/llm-vram-calculator tag:nodegrove.io,2026-10-22:/tools/llm-vram-calculator Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. Weights + KV cache + overhead for any model, with the formula shown. https://nodegrove.io/create/anima tag:nodegrove.io,2026-10-22:/create/anima Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. How much VRAM Anima needs in ComfyUI, which file fits each card size, and where to run it. https://nodegrove.io/create/flux-2 tag:nodegrove.io,2026-10-22:/create/flux-2 Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. How much VRAM FLUX.2 needs in ComfyUI, which file fits each card size, and where to run it. https://nodegrove.io/create/flux-3-image-klein tag:nodegrove.io,2026-10-22:/create/flux-3-image-klein Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. How much VRAM FLUX 3 Image Klein needs in ComfyUI, which file fits each card size, and where to run it. https://nodegrove.io/create/hunyuanvideo-1.5 tag:nodegrove.io,2026-10-22:/create/hunyuanvideo-1.5 Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. How much VRAM HunyuanVideo 1.5 needs in ComfyUI, which file fits each card size, and where to run it. https://nodegrove.io/create/krea-2 tag:nodegrove.io,2026-10-22:/create/krea-2 Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. How much VRAM Krea 2 needs in ComfyUI, which file fits each card size, and where to run it. https://nodegrove.io/create/ltx-2.5 tag:nodegrove.io,2026-10-22:/create/ltx-2.5 Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. How much VRAM LTX-2.5 needs in ComfyUI, which file fits each card size, and where to run it. https://nodegrove.io/create/minimax-h3 tag:nodegrove.io,2026-10-22:/create/minimax-h3 Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. How much VRAM MiniMax H3 needs in ComfyUI, which file fits each card size, and where to run it. https://nodegrove.io/create/qwen-image-2.1 tag:nodegrove.io,2026-10-22:/create/qwen-image-2.1 Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. How much VRAM Qwen-Image-2.1 needs in ComfyUI, which file fits each card size, and where to run it. https://nodegrove.io/create/qwen-image-edit-2511 tag:nodegrove.io,2026-10-22:/create/qwen-image-edit-2511 Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. How much VRAM Qwen-Image-Edit-2511 needs in ComfyUI, which file fits each card size, and where to run it. https://nodegrove.io/create/wan-2.2 tag:nodegrove.io,2026-10-22:/create/wan-2.2 Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. How much VRAM Wan 2.2 needs in ComfyUI, which file fits each card size, and where to run it. https://nodegrove.io/create/z-image tag:nodegrove.io,2026-10-22:/create/z-image Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. How much VRAM Z-Image needs in ComfyUI, which file fits each card size, and where to run it. https://nodegrove.io/gpus/a100-80 tag:nodegrove.io,2026-10-22:/gpus/a100-80 Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. What fits on an A100 at Q4 and Q8, how fast each model answers, how much context fits, and where the card stops. https://nodegrove.io/gpus/arc-b580 tag:nodegrove.io,2026-10-22:/gpus/arc-b580 Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. What fits on an Arc B580 at Q4 and Q8, how fast each model answers, how much context fits, and where the card stops. https://nodegrove.io/gpus/arc-pro-b60 tag:nodegrove.io,2026-10-22:/gpus/arc-pro-b60 Thu, 22 Oct 2026 00:00:00 GMT Updated 22 October 2026. What fits on an Arc Pro B60 at Q4 and Q8, how fast each model answers, how much context fits, and where the card stops. https://nodegrove.io/benchmarks tag:nodegrove.io,2026-10-20:/benchmarks#finding Tue, 20 Oct 2026 00:00:00 GMT In Nodegrove's runs of 20 October 2026, an RTX PRO 6000 finished three AI jobs 2.7 to 3.5 times faster than an RTX 6000 Ada and cost 35 to 50% less per output, at on-demand list prices. For LLMs, our memory estimate came within 1% of the peak we measured.