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.

Qwen3 Reranker

Rank passages by relevance to your query, on your device.

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

Score and reorder candidate passages by relevance to a query using Qwen3 0.6B and the original yes/no causal-language-model scoring prompt. Runs locally after model downloads; returns complete passage scores and original indices.

License
Apache-2.0
Model released
Jun 5, 2025
Source repository created
May 29, 2025
Model weights · FP16
1.2 GB

Best for

  • Rerank a short list of passages returned by semantic or lexical search
  • Explore local factual, programming and multilingual retrieval with human review

Keep in mind

  • Scores are query-dependent relevance signals, not calibrated probabilities, factuality checks or reliable accept/reject thresholds. Even unrelated passages receive scores.
  • Each document requires a separate forward pass; retrieve a shortlist first. Long text is right-truncated inside the prompt token budget, preserving the fixed system/assistant suffix; relevant text may be lost.
  • Default upstream retrieval instruction gives weak separation for the negation fixture. The saved sample and measured recommendation use an explicit direct-answer/negation-aware instruction; this does not guarantee general negation understanding.
  • Quality evidence covers four short authored factual/code/policy/Chinese tasks and an irrelevant stability case, not broad multilingual, domain, adversarial or safety accuracy.
  • Input instruction/query/documents can influence ranking through prompt injection. Do not rely on rankings alone for high-impact decisions.
  • No text generation, streaming, embeddings, unknown rejection or background network service. Browser ONNX quantization and native floating-point weights are independently tested.
  • OnlyFP16WebGPU passes the frozen suite onChrome/AppleM3Max. WASMFP16session allocation failed; Q4/Q8alternativeexports fail native-score agreement despite correcttop1. All failures retained.
  • Runtime uses pinnedONNXRuntimeWeb for the three-input FP16graph and pinnedTransformers.js only for tokenizer; cachedcausalLMgraphloader cannot execute this graph contract. Some shape operations run onCPU.
  • TheFP16weight file is about1.19GB; browser storage/CacheAPIavailability can prevent persistence even when inference works. Cache warnings are surfaced rather than treated as successful storage.

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
https://huggingface.co/Qwen/Qwen3-Reranker-0.6B
Release verification
Verified
Release evidence
Official June5 release announcement explicitly open-sources 0.6B/4B/8B embedding and reranking checkpoints. May29 repository creation/upload is recorded separately and not used as public release evidence.
Source updated
Apr 16, 2026
Browser model files
shawnw3i/Qwen3-Reranker-0.6B-ONNX
Files updated
May 22, 2026
Artifact revision
21fb0d00576b976132dbe2acb651dd9639881ecb

Input

QUERY → RANKED PASSAGES

Sample result

Qwen3 Reranker 0.6B

Settings

Return all passages with null, or1 through the number of documents. The numeric UI ceiling is a frontend policy. Smaller values hide lower-ranked passages but still score every candidate. Null returns the complete ranking.
Advanced settings 2
Complete prompt token budget, including fixed prefix and suffix; adapter truncates body text from the right. Browser adapter bound128–2048tokens controls full-logit activation memory. Lower values can remove relevant document content; higher values need more memory and inference time.
Task instruction inserted before the query and document; must be nonempty. Describe the relevance you need. Different instructions change scores and ranking; the recorded sample uses direct-answer and negation guidance.

Output

Find the most relevant passage

Add a query and candidate passages above. We’ll sort them from most to least relevant.

Benchmark evidence

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

Qwen3 Reranker 0.6B

51 configurations tested
Passed20
Failed31
TestedNov 2, 2026
RuntimeONNX Runtime Web + Transformers.js tokenizer 1.26.0-dev.20260416-b7804b056c / 4.2.0

Quality criteria

{ "version": 1, "expectedTop1Accuracy": 1, "minimumPositiveLogitMargin": 0.5, "maximumCpuScoreDifference": 0.08, "allScoresFinite": true, "minimumUsablePairsPerSecond": 0.25, "goodPairsPerSecond": 2, "scope": "Four short authored factual/code/negation/Chinese retrieval queries with hard negatives, plus all-irrelevant stability; not broad retrieval, multilingual or safety certification. All complete rankings retained; no calibrated relevance threshold or unknown rejection.", "browserMaximumTokens": 2048 }

PrecisionDeviceResult
FP16GPU · WebGPUAcceleratedPassedFast
FP16GPU · WebGPUAcceleratedPassedFast
FP16GPU · WebGPUAcceleratedPassedFast
FP16GPU · WebGPUAcceleratedPassedFast
FP16GPU · WebGPUAcceleratedPassedFast
FP16GPU · WebGPUAcceleratedPassedFast
FP16GPU · WebGPUAcceleratedPassedFast
FP16GPU · WebGPUAcceleratedPassedFast
FP16GPU · WebGPUAcceleratedPassedFast
Q4GPU · WebGPUAcceleratedPassedFast
Q4GPU · WebGPUAcceleratedPassedFast
FP16GPU · WebGPUAcceleratedPassedFast
Q8GPU · WebGPUAcceleratedPassedSlow
Q8GPU · WebGPUAcceleratedPassedSlow
Q8GPU · WebGPUAcceleratedPassedSlow
Q8CPU · WASMPassedModerate
Q8CPU · WASMPassedModerate
Q8CPU · WASMPassedModerate
Q4CPU · WASMPassedSlow
Q4CPU · WASMPassedSlow
FP16CPU · WASMFailed—
Q4CPU · WASMFailed—
Q4CPU · WASMFailed—
Q4CPU · WASMFailed—
Q4CPU · WASMFailed—
Q4CPU · WASMFailed—
Q4CPU · WASMFailed—
Q4CPU · WASMFailed—
Q4CPU · WASMFailed—
Q4GPU · WebGPUFailed—
Q4GPU · WebGPUFailed—
Q4GPU · WebGPUFailed—
Q4GPU · WebGPUFailed—
Q4GPU · WebGPUFailed—
Q4GPU · WebGPUFailed—
Q4GPU · WebGPUFailed—
Q4GPU · WebGPUFailed—
Q8CPU · WASMFailed—
Q8CPU · WASMFailed—
Q8CPU · WASMFailed—
Q8CPU · WASMFailed—
Q8CPU · WASMFailed—
Q8CPU · WASMFailed—
Q8CPU · WASMFailed—
Q8GPU · WebGPUFailed—
Q8GPU · WebGPUFailed—
Q8GPU · WebGPUFailed—
Q8GPU · WebGPUFailed—
Q8GPU · WebGPUFailed—
Q8GPU · WebGPUFailed—
Q8GPU · WebGPUFailed—
Test setup & sample details
Quality gate
{ "version": 1, "expectedTop1Accuracy": 1, "minimumPositiveLogitMargin": 0.5, "maximumCpuScoreDifference": 0.08, "allScoresFinite": true, "minimumUsablePairsPerSecond": 0.25, "goodPairsPerSecond": 2, "scope": "Four short authored factual/code/negation/Chinese retrieval queries with hard negatives, plus all-irrelevant stability; not broad retrieval, multilingual or safety certification. All complete rankings retained; no calibrated relevance threshold or unknown rejection.", "browserMaximumTokens": 2048 }
Tested
2026-11-02T18:09:59.530Z
Environment
{ "hardware": "Apple M3 Max", "os": "macOS27.2", "runtime": "Transformers.js4.2.0", "scope": "Isolated Chrome module workers; pinned local mirror loads" }
Runtime
ONNX Runtime Web + Transformers.js tokenizer 1.26.0-dev.20260416-b7804b056c / 4.2.0
Artifact revision
21fb0d00576b976132dbe2acb651dd9639881ecb
Fixture
{
  "file": "model-samples/qwen3-reranker-capital.json",
  "url": "https://assets.localai.tools/samples/v1/files/fcb30f2647fd37f102ebd90d61b0fd0bcb050ce208ff9caef38f79144fff9358/qwen3-reranker-capital.json",
  "size": 356,
  "sha1": "202d7afe4b252011e63de03681b2849227d4a46b",
  "sha256": "fcb30f2647fd37f102ebd90d61b0fd0bcb050ce208ff9caef38f79144fff9358"
}
Rerun when
Exact original/ONNX artifact revision or tokenizer; Pinned runtime or relevant browser/backend/OS behavior; Fixture bytes/expected ranking or quality gate; Prompt/truncation/yes-no scoring/preprocessing
Per-case quality & failure details
[
  {
    "precision": "fp16",
    "device": "wasm",
    "status": "fail",
    "quality": "Not embedded",
    "runtime": {
      "loadMs": null,
      "inferenceMs": null,
      "totalMs": 4142.109999999404,
      "grade": "failed",
      "engine": {
        "name": "ONNX Runtime Web + Transformers.js tokenizer",
        "version": "1.26.0-dev.20260416-b7804b056c / 4.2.0"
      },
      "acceleration": "cpu",
      "accelerationDetails": "ONNX Runtime WebGPU with CPU-assigned shape operations reported by runtime; CPUtokenizer/ranking. Not full GPU delegation."
    },
    "failure": "Error: Can't create a session. ERROR_CODE: 6, ERROR_MESSAGE: std::bad_alloc"
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 10.65625,
      "cpuScoreDifference": 0.002395129930311657,
      "passed": true
    },
    "runtime": {
      "loadMs": 5080.954999998212,
      "inferenceMs": 314.9150000065565,
      "totalMs": 5395.870000004768,
      "pairsPerSecond": 9.526380134123622,
      "grade": "good",
      "engine": {
        "name": "ONNX Runtime Web + Transformers.js tokenizer",
        "version": "1.26.0-dev.20260416-b7804b056c / 4.2.0"
      },
      "acceleration": "partial",
      "accelerationDetails": "ONNX Runtime WebGPU with CPU-assigned shape operations reported by runtime; CPUtokenizer/ranking. Not full GPU delegation."
    }
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 2.8671875,
      "cpuScoreDifference": 0.0036258450810306986,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 196.82999999821186,
      "totalMs": 196.82999999821186,
      "pairsPerSecond": 15.241579027725724,
      "grade": "excellent",
      "engine": {
        "name": "ONNX Runtime Web + Transformers.js tokenizer",
        "version": "1.26.0-dev.20260416-b7804b056c / 4.2.0"
      },
      "acceleration": "partial",
      "accelerationDetails": "ONNX Runtime WebGPU with CPU-assigned shape operations reported by runtime; CPUtokenizer/ranking. Not full GPU delegation."
    }
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 2.7578125,
      "cpuScoreDifference": 0.020679166305295538,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 184.46500000357628,
      "totalMs": 184.46500000357628,
      "pairsPerSecond": 16.263247770264485,
      "grade": "excellent",
      "engine": {
        "name": "ONNX Runtime Web + Transformers.js tokenizer",
        "version": "1.26.0-dev.20260416-b7804b056c / 4.2.0"
      },
      "acceleration": "partial",
      "accelerationDetails": "ONNX Runtime WebGPU with CPU-assigned shape operations reported by runtime; CPUtokenizer/ranking. Not full GPU delegation."
    }
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 12.515625,
      "cpuScoreDifference": 0.0005681657456095028,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 181.6650000065565,
      "totalMs": 181.6650000065565,
      "pairsPerSecond": 16.513912971082632,
      "grade": "excellent",
      "engine": {
        "name": "ONNX Runtime Web + Transformers.js tokenizer",
        "version": "1.26.0-dev.20260416-b7804b056c / 4.2.0"
      },
      "acceleration": "partial",
      "accelerationDetails": "ONNX Runtime WebGPU with CPU-assigned shape operations reported by runtime; CPUtokenizer/ranking. Not full GPU delegation."
    }
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": null,
      "positiveLogitMargin": null,
      "cpuScoreDifference": 0.000009399821355189384,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 174.60500000417233,
      "totalMs": 174.60500000417233,
      "pairsPerSecond": 17.18163855518635,
      "grade": "excellent",
      "engine": {
        "name": "ONNX Runtime Web + Transformers.js tokenizer",
        "version": "1.26.0-dev.20260416-b7804b056c / 4.2.0"
      },
      "acceleration": "partial",
      "accelerationDetails": "ONNX Runtime WebGPU with CPU-assigned shape operations reported by runtime; CPUtokenizer/ranking. Not full GPU delegation."
    }
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 10.65625,
      "cpuScoreDifference": 0.002395129930311657,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 180.75,
      "totalMs": 180.75,
      "pairsPerSecond": 16.597510373443985,
      "grade": "excellent",
      "engine": {
        "name": "ONNX Runtime Web + Transformers.js tokenizer",
        "version": "1.26.0-dev.20260416-b7804b056c / 4.2.0"
      },
      "acceleration": "partial",
      "accelerationDetails": "ONNX Runtime WebGPU with CPU-assigned shape operations reported by runtime; CPUtokenizer/ranking. Not full GPU delegation."
    }
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 10.65625,
      "cpuScoreDifference": 0.002395129930311657,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 181.40999999642372,
      "totalMs": 181.40999999642372,
      "pairsPerSecond": 16.53712584785371,
      "grade": "excellent",
      "engine": {
        "name": "ONNX Runtime Web + Transformers.js tokenizer",
        "version": "1.26.0-dev.20260416-b7804b056c / 4.2.0"
      },
      "acceleration": "partial",
      "accelerationDetails": "ONNX Runtime WebGPU with CPU-assigned shape operations reported by runtime; CPUtokenizer/ranking. Not full GPU delegation."
    }
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 10.65625,
      "cpuScoreDifference": 0.002395129930311657,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 180.34000000357628,
      "totalMs": 180.34000000357628,
      "pairsPerSecond": 16.63524453776482,
      "grade": "excellent",
      "engine": {
        "name": "ONNX Runtime Web + Transformers.js tokenizer",
        "version": "1.26.0-dev.20260416-b7804b056c / 4.2.0"
      },
      "acceleration": "partial",
      "accelerationDetails": "ONNX Runtime WebGPU with CPU-assigned shape operations reported by runtime; CPUtokenizer/ranking. Not full GPU delegation."
    }
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 10.65625,
      "cpuScoreDifference": 0.002395129930311657,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 176.42500001192093,
      "totalMs": 176.42500001192093,
      "pairsPerSecond": 17.004392800324737,
      "grade": "excellent",
      "engine": {
        "name": "ONNX Runtime Web + Transformers.js tokenizer",
        "version": "1.26.0-dev.20260416-b7804b056c / 4.2.0"
      },
      "acceleration": "partial",
      "accelerationDetails": "ONNX Runtime WebGPU with CPU-assigned shape operations reported by runtime; CPUtokenizer/ranking. Not full GPU delegation."
    }
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 10.65625,
      "cpuScoreDifference": 0.002395129930311657,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 180.6850000023842,
      "totalMs": 180.6850000023842,
      "pairsPerSecond": 16.603481196338457,
      "grade": "excellent",
      "engine": {
        "name": "ONNX Runtime Web + Transformers.js tokenizer",
        "version": "1.26.0-dev.20260416-b7804b056c / 4.2.0"
      },
      "acceleration": "partial",
      "accelerationDetails": "ONNX Runtime WebGPU with CPU-assigned shape operations reported by runtime; CPUtokenizer/ranking. Not full GPU delegation."
    }
  },
  {
    "precision": "q4",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.584003448486328,
      "cpuScoreDifference": 0.11164901595208176,
      "passed": false
    },
    "runtime": {
      "loadMs": 23190.340000003576,
      "inferenceMs": 3611.125,
      "totalMs": 26801.465000003576,
      "pairsPerSecond": 0.830766035515248,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q4",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 5.395444869995117,
      "cpuScoreDifference": 0.4155872787931898,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 3132.934999987483,
      "totalMs": 3132.934999987483,
      "pairsPerSecond": 0.9575685419620853,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q4",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 4.061840057373047,
      "cpuScoreDifference": 0.08335508050595608,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2798.265000000596,
      "totalMs": 2798.265000000596,
      "pairsPerSecond": 1.0720928861274257,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q4",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 11.408878326416016,
      "cpuScoreDifference": 0.002398360141037781,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2577.6500000059605,
      "totalMs": 2577.6500000059605,
      "pairsPerSecond": 1.163850794325476,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q4",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": null,
      "positiveLogitMargin": null,
      "cpuScoreDifference": 0.00002289330735190944,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2548.1299999952316,
      "totalMs": 2548.1299999952316,
      "pairsPerSecond": 1.1773339664795808,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q4",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.584003448486328,
      "cpuScoreDifference": 0.11164901595208176,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2489.314999997616,
      "totalMs": 2489.314999997616,
      "pairsPerSecond": 1.2051508145826757,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q4",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.584003448486328,
      "cpuScoreDifference": 0.11164901595208176,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2417.655000001192,
      "totalMs": 2417.655000001192,
      "pairsPerSecond": 1.24087183655175,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q4",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.584003448486328,
      "cpuScoreDifference": 0.11164901595208176,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2444.234999999404,
      "totalMs": 2444.234999999404,
      "pairsPerSecond": 1.2273778912423443,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q4",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.584003448486328,
      "cpuScoreDifference": 0.11164901595208176,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2427.130000010133,
      "totalMs": 2427.130000010133,
      "pairsPerSecond": 1.236027736457246,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q4",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.584003448486328,
      "cpuScoreDifference": 0.11164901595208176,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2418.7249999940395,
      "totalMs": 2418.7249999940395,
      "pairsPerSecond": 1.2403228973973448,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.584008693695068,
      "cpuScoreDifference": 0.11164829555449807,
      "passed": false
    },
    "runtime": {
      "loadMs": 24059.989999994636,
      "inferenceMs": 602.2550000101328,
      "totalMs": 24662.24500000477,
      "pairsPerSecond": 4.981278694157003,
      "grade": "good"
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 5.395452499389648,
      "cpuScoreDifference": 0.41558823130212474,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 301.0049999952316,
      "totalMs": 301.0049999952316,
      "pairsPerSecond": 9.966611850459376,
      "grade": "good"
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 4.061850547790527,
      "cpuScoreDifference": 0.08335493237827685,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 298.6600000113249,
      "totalMs": 298.6600000113249,
      "pairsPerSecond": 10.044867072544843,
      "grade": "excellent"
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 11.4088773727417,
      "cpuScoreDifference": 0.0023983768381609094,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 276.5700000077486,
      "totalMs": 276.5700000077486,
      "pairsPerSecond": 10.84716346644954,
      "grade": "excellent"
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": null,
      "positiveLogitMargin": null,
      "cpuScoreDifference": 0.000022893812156904004,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 274.85999999940395,
      "totalMs": 274.85999999940395,
      "pairsPerSecond": 10.914647456910812,
      "grade": "excellent"
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.584008693695068,
      "cpuScoreDifference": 0.11164829555449807,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 298.60999999940395,
      "totalMs": 298.60999999940395,
      "pairsPerSecond": 10.046549010434976,
      "grade": "excellent"
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.584008693695068,
      "cpuScoreDifference": 0.11164829555449807,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 306.09999999403954,
      "totalMs": 306.09999999403954,
      "pairsPerSecond": 9.800718719563596,
      "grade": "good"
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.584008693695068,
      "cpuScoreDifference": 0.11164829555449807,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 283.9450000077486,
      "totalMs": 283.9450000077486,
      "pairsPerSecond": 10.56542640271226,
      "grade": "excellent"
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.584008693695068,
      "cpuScoreDifference": 0.11164829555449807,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 271.4350000023842,
      "totalMs": 271.4350000023842,
      "pairsPerSecond": 11.05236981219684,
      "grade": "excellent"
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.584008693695068,
      "cpuScoreDifference": 0.11164829555449807,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 271.294999986887,
      "totalMs": 271.294999986887,
      "pairsPerSecond": 11.05807331556057,
      "grade": "excellent"
    }
  },
  {
    "precision": "q8",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.537567138671875,
      "cpuScoreDifference": 0.3185248074825542,
      "passed": false
    },
    "runtime": {
      "loadMs": 34700.92500001192,
      "inferenceMs": 1942.7950000017881,
      "totalMs": 36643.72000001371,
      "pairsPerSecond": 1.5441670376942698,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 5.734798431396484,
      "cpuScoreDifference": 0.45759459786360057,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2038.4549999982119,
      "totalMs": 2038.4549999982119,
      "pairsPerSecond": 1.4717028337650975,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 2.689192771911621,
      "cpuScoreDifference": 0.059303751069203936,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 1998.7549999952316,
      "totalMs": 1998.7549999952316,
      "pairsPerSecond": 1.500934331625015,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.921249389648438,
      "cpuScoreDifference": 0.01661645659227638,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 1824.300000011921,
      "totalMs": 1824.300000011921,
      "pairsPerSecond": 1.6444663706519742,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": null,
      "positiveLogitMargin": null,
      "cpuScoreDifference": 0.00012903131765596045,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 1812.684999987483,
      "totalMs": 1812.684999987483,
      "pairsPerSecond": 1.6550034893104515,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.537567138671875,
      "cpuScoreDifference": 0.3185248074825542,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2084.185000002384,
      "totalMs": 2084.185000002384,
      "pairsPerSecond": 1.4394115685491298,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.537567138671875,
      "cpuScoreDifference": 0.3185248074825542,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 1963.7300000041723,
      "totalMs": 1963.7300000041723,
      "pairsPerSecond": 1.5277049288820896,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.537567138671875,
      "cpuScoreDifference": 0.3185248074825542,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 1901.765000000596,
      "totalMs": 1901.765000000596,
      "pairsPerSecond": 1.5774819706951488,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.537567138671875,
      "cpuScoreDifference": 0.3185248074825542,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 1899.2099999934435,
      "totalMs": 1899.2099999934435,
      "pairsPerSecond": 1.5796041512051626,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 9.537567138671875,
      "cpuScoreDifference": 0.3185248074825542,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 1914.3700000047684,
      "totalMs": 1914.3700000047684,
      "pairsPerSecond": 1.5670951801336876,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 6.796820163726807,
      "cpuScoreDifference": 0.6276123601284331,
      "passed": false
    },
    "runtime": {
      "loadMs": 28875.92999999225,
      "inferenceMs": 2579.125,
      "totalMs": 31455.05499999225,
      "pairsPerSecond": 1.163185188775263,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 3.573981285095215,
      "cpuScoreDifference": 0.39709107674789335,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2474.765000000596,
      "totalMs": 2474.765000000596,
      "pairsPerSecond": 1.2122363133466318,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 1.6537389755249023,
      "cpuScoreDifference": 0.07699993560445373,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2483.5349999964237,
      "totalMs": 2483.5349999964237,
      "pairsPerSecond": 1.2079555955540469,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 6.187666893005371,
      "cpuScoreDifference": 0.0436560333540732,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2284.234999999404,
      "totalMs": 2284.234999999404,
      "pairsPerSecond": 1.3133499836929137,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": null,
      "positiveLogitMargin": null,
      "cpuScoreDifference": 0.0002044212635123203,
      "passed": true
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2293.7099999934435,
      "totalMs": 2293.7099999934435,
      "pairsPerSecond": 1.3079247158570941,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 6.796820163726807,
      "cpuScoreDifference": 0.6276123601284331,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2377.7549999952316,
      "totalMs": 2377.7549999952316,
      "pairsPerSecond": 1.261694329317367,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 6.796820163726807,
      "cpuScoreDifference": 0.6276123601284331,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2504.479999989271,
      "totalMs": 2504.479999989271,
      "pairsPerSecond": 1.1978534466287818,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 6.796820163726807,
      "cpuScoreDifference": 0.6276123601284331,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2375.8250000029802,
      "totalMs": 2375.8250000029802,
      "pairsPerSecond": 1.262719265937616,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 6.796820163726807,
      "cpuScoreDifference": 0.6276123601284331,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2424.4899999946356,
      "totalMs": 2424.4899999946356,
      "pairsPerSecond": 1.2373736332204455,
      "grade": "acceptable"
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "fail",
    "quality": {
      "finite": true,
      "correctTop": true,
      "top1Accuracy": 1,
      "positiveLogitMargin": 6.796820163726807,
      "cpuScoreDifference": 0.6276123601284331,
      "passed": false
    },
    "runtime": {
      "loadMs": 0,
      "inferenceMs": 2447.5750000029802,
      "totalMs": 2447.5750000029802,
      "pairsPerSecond": 1.2257029917352265,
      "grade": "acceptable"
    }
  }
]

Something not quite right?

Tell us what happened with Qwen3 Reranker. 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.