A model is still running

A model is still running or downloading. Leaving will stop the current operation. Stay on this page to let it finish.

Skip to content
LocalAI.toolsBeta

Join the waitlist

Get an email when the new apps and model runs launch.

No account needed.

SigLIP 2 Base · 224

Compare an image with your own category descriptions using SigLIP 2 at 224 × 224 resolution. Returns independent image–text matching scores.

Browser inferencePowered by WebAI · Private · Unlimited runs
About this model

Compare an image with your own category descriptions using SigLIP 2 at 224 × 224 resolution. Returns independent image–text matching scores.

Created by
Google
License
Apache-2.0
Model released
Feb 20, 2025
Source repository created
Apr 4, 2022
Model weights · Q8
378.0 MB

Best for

  • Zero-shot image categorization with custom labels.
  • Image-text matching and exploratory content organization.

Keep in mind

  • Images are resized to 224 × 224; small details or unusual domains may be misclassified.
  • Labels are truncated to the tokenizer context; scores are independent and do not sum to one.
  • The standard English fixture is a scoped sanity check, not broad multilingual accuracy validation.
  • Do not use matching scores as authoritative identity, medical, legal or safety judgments.
  • Browser quality evidence covers the supplied fixture only; this is not broad task validation.

Processed on your deviceInference runs in this browser. Input/output sharing requires consent for each run. Basic run logs include device information and approximate connection location. Model and runtime files download from external hosts. Saved downloads may be cleared by your browser.

Sources & technical details
Original model
google/siglip2-base-patch16-224
Release verification
Verified
Release evidence
Official Google release commit adds checkpoints and explicitly links SigLIP2 B/16 224 weights; original release, distinct from Feb21 Transformers artifact updates.
Source updated
May 19, 2025
Browser model files
onnx-community/siglip2-base-patch16-224-ONNX
Files updated
Feb 21, 2025
Artifact revision
ba1f3b0843f24bc5417d38e19c37b287d719b2f4

Input

IMAGE

Sample result

SigLIP 2 Base · 224

Settings

Adds a self-contained SVG presentation image alongside the unchanged native model fields. When enabled, draws model predictions or presents generated media and text. Adds CPU postprocessing time and image bytes. It does not change model predictions.
Advanced settings 1
English template containing one {} placeholder. Change the wording to suit your labels while retaining the required {} placeholder. Different wording changes scores and rankings; longer or more elaborate prompts are not necessarily better.

Output

Your result will appear here

Add input or try the sample, then run the model.

Benchmark evidence

Saved worker evidence is separate from benchmarks run on your device. It does not guarantee speed or quality on every browser.

SigLIP 2 Base · 224

4 configurations tested
Passed2
Failed2
TestedNov 4, 2026
RuntimeTransformers.js 4.2.0

Quality criteria

Finite scores in [0,1], unique labels, expected person label ranked first, and its score is at least 10 times the strongest negative.

PrecisionDeviceResult
Q8GPU · WebGPUAcceleratedPassedModerate
Q8CPU · WASMPassedModerate
FP16CPU · WASMFailed—
FP16GPU · WebGPUFailed—
Test setup & sample details
Quality gate
Finite scores in [0,1], unique labels, expected person label ranked first, and its score is at least 10 times the strongest negative.
Tested
2026-11-04T16:56:20.968Z
Environment
Not recorded
Runtime
Transformers.js 4.2.0
Artifact revision
ba1f3b0843f24bc5417d38e19c37b287d719b2f4
Fixture
{
  "file": "modnet-portrait-pexels-5965592.jpg",
  "url": "https://f005.backblazeb2.com/file/hubters-webai-public/assets/benchmarks/modnet-portrait-pexels-5965592.jpg",
  "size": 78124,
  "sha1": "38212739a4a4f0be25d84863cf991d7fb6c6d818",
  "width": 1024,
  "height": 683,
  "license": "Pexels License",
  "labels": {
    "file": "siglip-base-patch16-224.json",
    "url": "https://f005.backblazeb2.com/file/hubters-webai-public/assets/benchmarks/siglip-base-patch16-224.json",
    "size": 460,
    "sha1": "fecc3379e545c31c5fc0dbcd8ca8777189f661e1"
  }
}
Rerun when
artifact or tokenizer revision change; runtime version change; fixture, labels, or quality-gate change; browser/backend behavior change
Per-case quality & failure details
[
  {
    "precision": "q8",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "finite": true,
      "unique": true,
      "topLabel": "a portrait photo of a person",
      "topToNegativeRatio": 24.848950108462567,
      "passed": true,
      "visualization": true
    },
    "runtime": {
      "loadMs": 2752.0999999940395,
      "inferenceMs": 1528.2000000178814,
      "totalMs": 5549.9000000059605,
      "grade": "good",
      "imagesPerSecond": 0.6543646119541284,
      "artifactRevision": "ba1f3b0843f24bc5417d38e19c37b287d719b2f4",
      "engine": {
        "name": "Transformers.js",
        "version": "4.2.0"
      }
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "unique": true,
      "topLabel": "a portrait photo of a person",
      "topToNegativeRatio": 63.35618606026562,
      "passed": true,
      "visualization": true
    },
    "runtime": {
      "loadMs": 2785.199999988079,
      "inferenceMs": 1733.699999988079,
      "totalMs": 5857.9000000059605,
      "grade": "good",
      "imagesPerSecond": 0.5768010613179189,
      "artifactRevision": "ba1f3b0843f24bc5417d38e19c37b287d719b2f4",
      "engine": {
        "name": "Transformers.js",
        "version": "4.2.0"
      }
    }
  },
  {
    "precision": "fp16",
    "device": "wasm",
    "status": "fail",
    "quality": "Not embedded",
    "runtime": {
      "loadMs": null,
      "inferenceMs": null,
      "totalMs": 4723,
      "grade": "failed"
    },
    "failure": "Can't create a session. ERROR_CODE: 1, ERROR_MESSAGE: /mnt/vss/_work/1/s/onnxruntime/core/graph/graph_utils.cc:30 int onnxruntime::graph_utils::GetIndexFromName(const Node &, const std::string &, bool) itr != node_args.end() was false. Attempting to get index by a name which does not exist:InsertedPrecisionFreeCast_/vision_model/encoder/layers.0/layer_norm1/Constant_output_0for node: /vision_model/encoder/layers.0/layer_norm1/Mul/SimplifiedLayerNormFusion/\n"
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "unique": true,
      "topLabel": "a portrait photo of a person",
      "topToNegativeRatio": 4.696770068504018,
      "passed": false,
      "visualization": true
    },
    "runtime": {
      "loadMs": 3273.2999999821186,
      "inferenceMs": 1148.699999988079,
      "totalMs": 5907.100000023842,
      "grade": "good",
      "imagesPerSecond": 0.8705493166278208,
      "artifactRevision": "ba1f3b0843f24bc5417d38e19c37b287d719b2f4",
      "engine": {
        "name": "Transformers.js",
        "version": "4.2.0"
      }
    },
    "failure": "Quality or 15-second interactive usability gate failed"
  }
]

SigLIP 2 Base · 384

4 configurations tested
Passed2
Failed2
TestedNov 4, 2026
RuntimeTransformers.js 4.2.0

Quality criteria

Finite scores in [0,1], unique labels, expected person label ranked first, and its score is at least 10 times the strongest negative.

PrecisionDeviceResult
Q8GPU · WebGPUAcceleratedPassedSlow
Q8CPU · WASMPassedSlow
FP16CPU · WASMFailed—
FP16GPU · WebGPUFailed—
Test setup & sample details
Quality gate
Finite scores in [0,1], unique labels, expected person label ranked first, and its score is at least 10 times the strongest negative.
Tested
2026-11-04T16:56:36.567Z
Environment
Not recorded
Runtime
Transformers.js 4.2.0
Artifact revision
99f5d40afb5331b81eecee4bb02c360fbb9cdf8a
Fixture
{
  "file": "modnet-portrait-pexels-5965592.jpg",
  "url": "https://f005.backblazeb2.com/file/hubters-webai-public/assets/benchmarks/modnet-portrait-pexels-5965592.jpg",
  "size": 78124,
  "sha1": "38212739a4a4f0be25d84863cf991d7fb6c6d818",
  "width": 1024,
  "height": 683,
  "license": "Pexels License",
  "labels": {
    "file": "siglip-base-patch16-224.json",
    "url": "https://f005.backblazeb2.com/file/hubters-webai-public/assets/benchmarks/siglip-base-patch16-224.json",
    "size": 460,
    "sha1": "fecc3379e545c31c5fc0dbcd8ca8777189f661e1"
  }
}
Rerun when
artifact or tokenizer revision change; runtime version change; fixture, labels, or quality-gate change; browser/backend behavior change
Per-case quality & failure details
[
  {
    "precision": "q8",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "finite": true,
      "unique": true,
      "topLabel": "a portrait photo of a person",
      "topToNegativeRatio": 82.04298595016047,
      "passed": true,
      "visualization": true
    },
    "runtime": {
      "loadMs": 2808.2999999821186,
      "inferenceMs": 3141.9000000059605,
      "totalMs": 7285,
      "grade": "good",
      "imagesPerSecond": 0.318278748527357,
      "artifactRevision": "99f5d40afb5331b81eecee4bb02c360fbb9cdf8a",
      "engine": {
        "name": "Transformers.js",
        "version": "4.2.0"
      }
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "unique": true,
      "topLabel": "a portrait photo of a person",
      "topToNegativeRatio": 92.46376425197458,
      "passed": true,
      "visualization": true
    },
    "runtime": {
      "loadMs": 2813.2000000178814,
      "inferenceMs": 3090.800000011921,
      "totalMs": 7300.5999999940395,
      "grade": "good",
      "imagesPerSecond": 0.3235408308516058,
      "artifactRevision": "99f5d40afb5331b81eecee4bb02c360fbb9cdf8a",
      "engine": {
        "name": "Transformers.js",
        "version": "4.2.0"
      }
    }
  },
  {
    "precision": "fp16",
    "device": "wasm",
    "status": "fail",
    "quality": "Not embedded",
    "runtime": {
      "loadMs": null,
      "inferenceMs": null,
      "totalMs": 4901.800000011921,
      "grade": "failed"
    },
    "failure": "Can't create a session. ERROR_CODE: 1, ERROR_MESSAGE: /mnt/vss/_work/1/s/onnxruntime/core/graph/graph_utils.cc:30 int onnxruntime::graph_utils::GetIndexFromName(const Node &, const std::string &, bool) itr != node_args.end() was false. Attempting to get index by a name which does not exist:InsertedPrecisionFreeCast_/vision_model/encoder/layers.0/layer_norm1/Constant_output_0for node: /vision_model/encoder/layers.0/layer_norm1/Mul/SimplifiedLayerNormFusion/\n"
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "unique": true,
      "topLabel": "a portrait photo of a person",
      "topToNegativeRatio": 4.144884739258839,
      "passed": false,
      "visualization": true
    },
    "runtime": {
      "loadMs": 3421.399999976158,
      "inferenceMs": 347.2000000178814,
      "totalMs": 5259.0999999940395,
      "grade": "excellent",
      "imagesPerSecond": 2.8801843316489006,
      "artifactRevision": "99f5d40afb5331b81eecee4bb02c360fbb9cdf8a",
      "engine": {
        "name": "Transformers.js",
        "version": "4.2.0"
      }
    },
    "failure": "Quality or 15-second interactive usability gate failed"
  }
]

SigLIP 2 Base · 512

4 configurations tested
Passed2
Failed2
TestedNov 4, 2026
RuntimeTransformers.js 4.2.0

Quality criteria

Finite scores in [0,1], unique labels, expected person label ranked first, and its score is at least 10 times the strongest negative.

PrecisionDeviceResult
Q8GPU · WebGPUAcceleratedPassedSlow
Q8CPU · WASMPassedVery slow
FP16CPU · WASMFailed—
FP16GPU · WebGPUFailed—
Test setup & sample details
Quality gate
Finite scores in [0,1], unique labels, expected person label ranked first, and its score is at least 10 times the strongest negative.
Tested
2026-11-04T16:56:55.977Z
Environment
Not recorded
Runtime
Transformers.js 4.2.0
Artifact revision
323fe16b35d8ced8182f8d9ff5e5ec935dd79bd7
Fixture
{
  "file": "modnet-portrait-pexels-5965592.jpg",
  "url": "https://f005.backblazeb2.com/file/hubters-webai-public/assets/benchmarks/modnet-portrait-pexels-5965592.jpg",
  "size": 78124,
  "sha1": "38212739a4a4f0be25d84863cf991d7fb6c6d818",
  "width": 1024,
  "height": 683,
  "license": "Pexels License",
  "labels": {
    "file": "siglip-base-patch16-224.json",
    "url": "https://f005.backblazeb2.com/file/hubters-webai-public/assets/benchmarks/siglip-base-patch16-224.json",
    "size": 460,
    "sha1": "fecc3379e545c31c5fc0dbcd8ca8777189f661e1"
  }
}
Rerun when
artifact or tokenizer revision change; runtime version change; fixture, labels, or quality-gate change; browser/backend behavior change
Per-case quality & failure details
[
  {
    "precision": "q8",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "finite": true,
      "unique": true,
      "topLabel": "a portrait photo of a person",
      "topToNegativeRatio": 112.37377035866375,
      "passed": true,
      "visualization": true
    },
    "runtime": {
      "loadMs": 2773.9000000059605,
      "inferenceMs": 5356,
      "totalMs": 9471.600000023842,
      "grade": "slow",
      "imagesPerSecond": 0.18670649738610903,
      "artifactRevision": "323fe16b35d8ced8182f8d9ff5e5ec935dd79bd7",
      "engine": {
        "name": "Transformers.js",
        "version": "4.2.0"
      }
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "unique": true,
      "topLabel": "a portrait photo of a person",
      "topToNegativeRatio": 94.59358610113073,
      "passed": true,
      "visualization": true
    },
    "runtime": {
      "loadMs": 2841.9000000059605,
      "inferenceMs": 4760.800000011921,
      "totalMs": 8925.300000011921,
      "grade": "good",
      "imagesPerSecond": 0.21004873130513696,
      "artifactRevision": "323fe16b35d8ced8182f8d9ff5e5ec935dd79bd7",
      "engine": {
        "name": "Transformers.js",
        "version": "4.2.0"
      }
    }
  },
  {
    "precision": "fp16",
    "device": "wasm",
    "status": "fail",
    "quality": "Not embedded",
    "runtime": {
      "loadMs": null,
      "inferenceMs": null,
      "totalMs": 5183.300000011921,
      "grade": "failed"
    },
    "failure": "Can't create a session. ERROR_CODE: 1, ERROR_MESSAGE: /mnt/vss/_work/1/s/onnxruntime/core/graph/graph_utils.cc:30 int onnxruntime::graph_utils::GetIndexFromName(const Node &, const std::string &, bool) itr != node_args.end() was false. Attempting to get index by a name which does not exist:InsertedPrecisionFreeCast_/vision_model/encoder/layers.0/layer_norm1/Constant_output_0for node: /vision_model/encoder/layers.0/layer_norm1/Mul/SimplifiedLayerNormFusion/\n"
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "unique": true,
      "topLabel": "a portrait photo of a person",
      "topToNegativeRatio": 2.634648996516492,
      "passed": false,
      "visualization": true
    },
    "runtime": {
      "loadMs": 3618.4000000059605,
      "inferenceMs": 273.5,
      "totalMs": 5400.9000000059605,
      "grade": "excellent",
      "imagesPerSecond": 3.656307129798903,
      "artifactRevision": "323fe16b35d8ced8182f8d9ff5e5ec935dd79bd7",
      "engine": {
        "name": "Transformers.js",
        "version": "4.2.0"
      }
    },
    "failure": "Quality or 15-second interactive usability gate failed"
  }
]
Unavailable checkpoints (2)

These checkpoints have not passed the required browser checks.

Base ·NaFlex · Unavailable in the browser; requires isolated WebGPU/WASM inference, quality, memory and public WebAI checks. Different preprocessing and architecture; not the requested fixed224checkpoint.

So400m ·256 · Unavailable in the browser; requires isolated WebGPU/WASM inference, quality, memory and public WebAI checks. Heavier official checkpoint; independent performance/memory evidence required.

Something not quite right?

Tell us what happened with SigLIP 2. Loading, results, or the page itself—we’re here for all of it.

Attachments Optional

Up to 3 files · 50 MB each

Browser, OS, CPU/GPU, memory, screen, storage, and network details are included automatically. Model inputs, results, and browsing history are not collected.