<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <id>tag:modelsthatrun.dev,2026-11-02:changes</id>
  <title>ModelsThatRun: changes</title>
  <subtitle>New models, sizes, runtimes and re-measured numbers, from our own runs on an RTX 4090.</subtitle>
  <link rel="self" type="application/atom+xml" href="https://modelsthatrun.dev/feed.xml"/>
  <link rel="alternate" type="text/html" href="https://modelsthatrun.dev/changes/"/>
  <updated>2026-11-14T12:44:57.210Z</updated>
  <author><name>ModelsThatRun</name></author>
  <entry>
    <id>tag:modelsthatrun.dev,2026-11-02:changes:2026-11-14</id>
    <title>14 Nov 2026: CPU-only rows for 44 models</title>
    <link rel="alternate" type="text/html" href="https://modelsthatrun.dev/changes/#2026-11-14"/>
    <updated>2026-11-14T12:44:57.210Z</updated>
    <content type="html">&lt;ul&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/ben2/&quot;&gt;BEN2&lt;/a&gt;: CPU-only rows (PyTorch and ONNX Runtime): base; no number for base on ONNX Runtime&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/birefnet/&quot;&gt;BiRefNet&lt;/a&gt;: CPU-only rows (PyTorch): general, hr, and lite&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/d-fine/&quot;&gt;D-FINE&lt;/a&gt;: CPU-only rows (PyTorch and ONNX Runtime): n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/deimv2/&quot;&gt;DEIMv2&lt;/a&gt;: CPU-only rows (PyTorch): s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/depth-anything-3-metric/&quot;&gt;Depth Anything 3 Metric&lt;/a&gt;: CPU-only row (PyTorch): large, without a number&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/depth-anything-3-mono/&quot;&gt;Depth Anything 3 Mono&lt;/a&gt;: CPU-only row (PyTorch): large, without a number&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/depth-anything-v2/&quot;&gt;Depth Anything V2&lt;/a&gt;: CPU-only rows (PyTorch): small, base, and large&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/eomt/&quot;&gt;EoMT&lt;/a&gt;: CPU-only rows (PyTorch): large-640 and large-1280&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/flux-2-klein/&quot;&gt;FLUX.2 [klein]&lt;/a&gt;: CPU-only rows (PyTorch): 4b and base-4b; no number for base-4b on PyTorch&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/gemma-4/&quot;&gt;Gemma 4&lt;/a&gt;: CPU-only rows (llama.cpp): 12b-q4_0 and 31b-q4_0&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/grounding-dino/&quot;&gt;Grounding DINO&lt;/a&gt;: CPU-only rows (PyTorch): tiny and base&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/gtr/&quot;&gt;GTR&lt;/a&gt;: CPU-only rows (PyTorch): s, m, l, and x, without a number&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/gtr-pose/&quot;&gt;GTR-Pose&lt;/a&gt;: CPU-only rows (PyTorch): s, m, l, and x, without a number&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/gtr-seg/&quot;&gt;GTR-Seg&lt;/a&gt;: CPU-only rows (PyTorch): s, m, l, and x, without a number&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/llmdet/&quot;&gt;LLMDet&lt;/a&gt;: CPU-only rows (PyTorch): tiny, base, and large&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/lt-detr/&quot;&gt;LT-DETR&lt;/a&gt;: CPU-only rows (PyTorch and ONNX Runtime): s, m, and l&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/lumina-image-2/&quot;&gt;Lumina-Image 2.0&lt;/a&gt;: CPU-only row (PyTorch): 2.0, without a number&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/mm-grounding-dino/&quot;&gt;MM-Grounding-DINO&lt;/a&gt;: CPU-only rows (PyTorch): tiny, base, and large&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/moge-2/&quot;&gt;MoGe-2&lt;/a&gt;: CPU-only rows (PyTorch): vits, vitb, and vitl&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/pixart-sigma/&quot;&gt;PixArt-Σ&lt;/a&gt;: CPU-only row (PyTorch): xl-2-1024-ms, without a number&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/playground-v2-5/&quot;&gt;Playground v2.5&lt;/a&gt;: CPU-only row (PyTorch): 1024px-aesthetic, without a number&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/qwen3/&quot;&gt;Qwen3&lt;/a&gt;: CPU-only rows (llama.cpp): 0.6b-q8_0 and 8b-q4_k_m&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/qwen3-5/&quot;&gt;Qwen3.5&lt;/a&gt;: CPU-only row (llama.cpp): 9b-q4_k_m&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/qwen3-8/&quot;&gt;Qwen3.8&lt;/a&gt;: CPU-only row (llama.cpp): 27b-q4_k_m&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/rf-detr/&quot;&gt;RF-DETR&lt;/a&gt;: CPU-only rows (PyTorch): nano, small, medium, and large&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/rf-detr-keypoint/&quot;&gt;RF-DETR Keypoint&lt;/a&gt;: CPU-only row (PyTorch): preview&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/rf-detr-seg/&quot;&gt;RF-DETR-Seg&lt;/a&gt;: CPU-only rows (PyTorch): nano, small, medium, large, xlarge, and 2xlarge&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/rt-detrv4/&quot;&gt;RT-DETRv4&lt;/a&gt;: CPU-only rows (PyTorch and ONNX Runtime): s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/sana-1-5/&quot;&gt;Sana 1.5&lt;/a&gt;: CPU-only rows (PyTorch): 1.6b and 4.8b, without a number&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/sana-sprint/&quot;&gt;Sana-Sprint&lt;/a&gt;: CPU-only row (PyTorch): 1.6b&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/sdxl-lightning/&quot;&gt;SDXL-Lightning&lt;/a&gt;: CPU-only row (PyTorch): 4step&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/sdxl-turbo/&quot;&gt;SDXL-Turbo&lt;/a&gt;: CPU-only row (PyTorch): 1.0&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/stable-diffusion-1-5/&quot;&gt;Stable Diffusion 1.5&lt;/a&gt;: CPU-only row (PyTorch): v1-5&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/sdxl/&quot;&gt;Stable Diffusion XL&lt;/a&gt;: CPU-only row (PyTorch): base, without a number&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolo11/&quot;&gt;YOLO11&lt;/a&gt;: CPU-only rows (PyTorch and ONNX Runtime): n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolo11-pose/&quot;&gt;YOLO11-pose&lt;/a&gt;: CPU-only rows (PyTorch and ONNX Runtime): n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolo11-seg/&quot;&gt;YOLO11-seg&lt;/a&gt;: CPU-only rows (PyTorch and ONNX Runtime): n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolo26/&quot;&gt;YOLO26&lt;/a&gt;: CPU-only rows (PyTorch and ONNX Runtime): n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolo26-pose/&quot;&gt;YOLO26-pose&lt;/a&gt;: CPU-only rows (PyTorch and ONNX Runtime): n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolo26-seg/&quot;&gt;YOLO26-seg&lt;/a&gt;: CPU-only rows (PyTorch and ONNX Runtime): n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolov8/&quot;&gt;YOLOv8&lt;/a&gt;: CPU-only rows (PyTorch and ONNX Runtime): n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolov8-pose/&quot;&gt;YOLOv8-pose&lt;/a&gt;: CPU-only rows (PyTorch and ONNX Runtime): n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolov8-seg/&quot;&gt;YOLOv8-seg&lt;/a&gt;: CPU-only rows (PyTorch and ONNX Runtime): n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/z-image/&quot;&gt;Z-Image&lt;/a&gt;: CPU-only rows (PyTorch): turbo and base, without a number&lt;/li&gt;&lt;/ul&gt;</content>
  </entry>
  <entry>
    <id>tag:modelsthatrun.dev,2026-11-02:changes:2026-11-13</id>
    <title>13 Nov 2026: First scores for FLUX.2 [klein]</title>
    <link rel="alternate" type="text/html" href="https://modelsthatrun.dev/changes/#2026-11-13"/>
    <updated>2026-11-13T06:33:59.987Z</updated>
    <content type="html">&lt;ul&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/flux-2-klein/&quot;&gt;FLUX.2 [klein]&lt;/a&gt;: First GenEval score: base-4b&lt;/li&gt;&lt;/ul&gt;</content>
  </entry>
  <entry>
    <id>tag:modelsthatrun.dev,2026-11-02:changes:2026-11-12</id>
    <title>12 Nov 2026: 8 new models; new sizes of FLUX.2 [klein] and Z-Image</title>
    <link rel="alternate" type="text/html" href="https://modelsthatrun.dev/changes/#2026-11-12"/>
    <updated>2026-11-12T13:11:27.685Z</updated>
    <content type="html">&lt;ul&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/lumina-image-2/&quot;&gt;Lumina-Image 2.0&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/pixart-sigma/&quot;&gt;PixArt-Σ&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/playground-v2-5/&quot;&gt;Playground v2.5&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/sana-1-5/&quot;&gt;Sana 1.5&lt;/a&gt;: Added, with sizes 1.6b and 4.8b&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/sana-sprint/&quot;&gt;Sana-Sprint&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/sdxl-lightning/&quot;&gt;SDXL-Lightning&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/sdxl-turbo/&quot;&gt;SDXL-Turbo&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/stable-diffusion-1-5/&quot;&gt;Stable Diffusion 1.5&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/flux-2-klein/&quot;&gt;FLUX.2 [klein]&lt;/a&gt;: New size: base-4b&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/z-image/&quot;&gt;Z-Image&lt;/a&gt;: New size: base&lt;/li&gt;&lt;/ul&gt;</content>
  </entry>
  <entry>
    <id>tag:modelsthatrun.dev,2026-11-02:changes:2026-11-11</id>
    <title>11 Nov 2026: TensorRT rows for D-FINE; new numbers for D-FINE</title>
    <link rel="alternate" type="text/html" href="https://modelsthatrun.dev/changes/#2026-11-11"/>
    <updated>2026-11-11T01:03:29.977Z</updated>
    <content type="html">&lt;ul&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/d-fine/&quot;&gt;D-FINE&lt;/a&gt;: TensorRT 10.13.2.6 rows: s, m, and l; n re-measured · RTX 4090: 10.4 ms (TensorRT 10.16.1.11 fp16, 640x640) → 10.5 ms (TensorRT 10.13.2.6 fp16, 640x640); n re-scored · box AP (TensorRT): 42.5 → 42.6&lt;/li&gt;&lt;/ul&gt;</content>
  </entry>
  <entry>
    <id>tag:modelsthatrun.dev,2026-11-02:changes:2026-11-10</id>
    <title>10 Nov 2026: 4 new models; TensorRT rows for D-FINE and YOLO26</title>
    <link rel="alternate" type="text/html" href="https://modelsthatrun.dev/changes/#2026-11-10"/>
    <updated>2026-11-10T21:17:17.633Z</updated>
    <content type="html">&lt;ul&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/flux-2-klein/&quot;&gt;FLUX.2 [klein]&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/lt-detr/&quot;&gt;LT-DETR&lt;/a&gt;: Added, with sizes s, m, and l&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/sdxl/&quot;&gt;Stable Diffusion XL&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/z-image/&quot;&gt;Z-Image&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/d-fine/&quot;&gt;D-FINE&lt;/a&gt;: TensorRT 10.16.1.11 row: n&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolo26/&quot;&gt;YOLO26&lt;/a&gt;: TensorRT 10.16.1.11 rows: s, m, l, and x&lt;/li&gt;&lt;/ul&gt;</content>
  </entry>
  <entry>
    <id>tag:modelsthatrun.dev,2026-11-02:changes:2026-11-03</id>
    <title>3 Nov 2026: 11 new models; new sizes of Depth Anything V2; new numbers for Qwen3 and RF-DETR</title>
    <link rel="alternate" type="text/html" href="https://modelsthatrun.dev/changes/#2026-11-03"/>
    <updated>2026-11-03T21:23:23.362Z</updated>
    <content type="html">&lt;ul&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/depth-anything-3-metric/&quot;&gt;Depth Anything 3 Metric&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/depth-anything-3-mono/&quot;&gt;Depth Anything 3 Mono&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/gemma-4/&quot;&gt;Gemma 4&lt;/a&gt;: Added, with sizes 12b-q4_0 and 31b-q4_0&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/qwen3-5/&quot;&gt;Qwen3.5&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/qwen3-8/&quot;&gt;Qwen3.8&lt;/a&gt;: Added&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolo11/&quot;&gt;YOLO11&lt;/a&gt;: Added, with sizes n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolo11-pose/&quot;&gt;YOLO11-pose&lt;/a&gt;: Added, with sizes n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolo11-seg/&quot;&gt;YOLO11-seg&lt;/a&gt;: Added, with sizes n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolov8/&quot;&gt;YOLOv8&lt;/a&gt;: Added, with sizes n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolov8-pose/&quot;&gt;YOLOv8-pose&lt;/a&gt;: Added, with sizes n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/yolov8-seg/&quot;&gt;YOLOv8-seg&lt;/a&gt;: Added, with sizes n, s, m, l, and x&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/depth-anything-v2/&quot;&gt;Depth Anything V2&lt;/a&gt;: New sizes: base and large&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/qwen3/&quot;&gt;Qwen3&lt;/a&gt;: 0.6b-q8_0 re-scored · GSM8K: 43.1 → 43.3; 8b-q4_k_m re-scored · GSM8K: 86.3 → 89.6&lt;/li&gt;&lt;li&gt;&lt;a href=&quot;https://modelsthatrun.dev/models/rf-detr/&quot;&gt;RF-DETR&lt;/a&gt;: nano re-measured · RTX 4090 · PyTorch fp16, 384x384: 9.9 → 9.8 ms&lt;/li&gt;&lt;/ul&gt;</content>
  </entry>
  <entry>
    <id>tag:modelsthatrun.dev,2026-11-02:changes:2026-11-02</id>
    <title>2 Nov 2026: Site launched with 21 models</title>
    <link rel="alternate" type="text/html" href="https://modelsthatrun.dev/changes/#2026-11-02"/>
    <updated>2026-11-02T00:00:00.000Z</updated>
    <content type="html">&lt;ul&gt;&lt;li&gt;Site launched with 21 models, each run and measured on our RTX 4090. &lt;a href=&quot;https://modelsthatrun.dev/models/&quot;&gt;See every model&lt;/a&gt;&lt;/li&gt;&lt;/ul&gt;</content>
  </entry>
</feed>
