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.

Gemma 4 E2B Instruct — Text, Image & Audio

Chat with text, ask about still images, and transcribe short WAV recordings locally with the complete Gemma4 E2B checkpoint. Language/vision use WebGPU; audio encoding uses CPU.

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

Chat with text, ask about still images, and transcribe short WAV recordings locally with the complete Gemma4 E2B checkpoint. Language/vision use WebGPU; audio encoding uses CPU.

Created by
Google
License
Apache-2.0
Model released
Mar 31, 2026
Source repository created
Mar 2, 2026
Model weights · Q4
2.6 GB

Best for

  • Local conversational assistants, drafting, rewriting, and summarization.
  • Browser-side experiments where prompts and generated text should stay on the device after the model download.

Keep in mind

  • Generated text can be incorrect, biased, unsafe, or fabricated and must be reviewed before consequential use.
  • LiteRT-LM's current web model support is limited and this worker requires WebGPU; WASM-only inference is not advertised.
  • The complete multimodal artifact is large and may exceed browser WASM/GPU memory budgets, especially for E4B. Only measured configurations are supported.
  • Video and sampled video-frame inputs are not supported. Use a still image instead.

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/gemma-4-E2B-it
Release verification
Verified
Release evidence
Gemma release log explicitly lists initial E2B/E4B checkpoints on March 31; excludes later MTP release and LiteRT conversion.
Source updated
Jul 20, 2026
Browser model files
litert-community/gemma-4-E2B-it-litert-lm
Files updated
Nov 5, 2026
Artifact revision
ec14a83775e46685aad47bff6bbb647f6c3d1c4f

Input

CHAT & TEXT
Record a clip, then run inference. This does not send live audio to the model.

Sample result

Gemma 4 E2B Instruct — Text, Image & Audio

Settings

Limits the number of newly generated output tokens. Lower values finish sooner but can cut answers short. Higher values allow longer answers and can use more time and memory; they do not guarantee better answers.
Limits how many ranked predictions or passages are returned. The listed range is a frontend selection policy, not a claimed model or runtime maximum; the adapter may accept values outside it. Lower values show fewer top results. Higher values include weaker matches and more output. Automatic/null returns all results when offered; it does not mean the model is more accurate.
Advanced settings 5
Adjusts randomness during token sampling. Lower values favor likely tokens and more repeatable wording. Higher values increase variation and can increase mistakes. It affects sampling only; some runtimes use zero for greedy decoding.
Chooses sampled rather than greedy token selection. On allows varied responses and makes sampling controls relevant. Off favors the most likely tokens for more repeatable responses; it does not guarantee correctness.
Restricts sampling to a set of tokens covering a cumulative probability mass. Lower values restrict word choice to more likely tokens. Higher values allow more varied but potentially less reliable wording. It affects sampling only; keep the default unless tuning generation.
Sets the pseudorandom seed used for generation. The listed range is a frontend selection policy, not a claimed model or runtime maximum; the adapter may accept values outside it. The same seed can help repeat sampled runs under the same runtime and settings, but does not guarantee identical results across devices. A different number changes the random sequence; larger numbers are not better.
Include the input image and answer as a self-contained SVG. Turn off to return only answer text and avoid presentation processing.
Estimated speedNot benchmarked

Output

Start a conversation

Send a message or try a sample. Replies appear here.

Benchmark evidence

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

Gemma 4 E2B Instruct — Text, Image & Audio

9 configurations tested
Passed9
Failed0
TestedNov 5, 2026
RuntimeLiteRT-LM JS 0.19.0

Quality criteria

{ "version": "multimodal-v1", "scope": "Text, image recognition/OCR/counting, speech and silence fixtures; per-fixture quality and genuine streaming parity." }

PrecisionDeviceResult
Q4GPU · WebGPUAcceleratedPassedModerate
Q4GPU · WebGPUAcceleratedPassedModerate
Q4GPU · WebGPUAcceleratedPassedModerate
Q4GPU · WebGPUAcceleratedPassedModerate
Q4GPU · WebGPUAcceleratedPassedModerate
Q4GPU · WebGPUAcceleratedPassedSlow
Q4GPU · WebGPUAcceleratedPassedVery slow
Q4GPU · WebGPUAcceleratedPassedVery slow
Q4GPU · WebGPUAcceleratedPassedVery slow
Test setup & sample details
Quality gate
{ "version": "multimodal-v1", "scope": "Text, image recognition/OCR/counting, speech and silence fixtures; per-fixture quality and genuine streaming parity." }
Tested
2026-11-05T17:34:22.031Z
Environment
Codex in-app Chromium on macOS; LiteRT-LM JS 0.19.0; WebGPU
Runtime
LiteRT-LM JS 0.19.0
Artifact revision
ec14a83775e46685aad47bff6bbb647f6c3d1c4f
Fixture
{
  "file": "model-samples/multimodal/gemma-4-e2b-it-litert-lm-text-fixture.json",
  "url": "https://assets.localai.tools/samples/v1/files/b44d7be442f7c7f5c4948ad305192078d7a23a7e0210e1afe93c67461f6f066a/gemma-4-e2b-it-litert-lm-text-fixture.json",
  "size": 431,
  "sha1": "e617fd367ad31a5b536e4685e09de1f92f5e7e01",
  "sha256": "b44d7be442f7c7f5c4948ad305192078d7a23a7e0210e1afe93c67461f6f066a"
}
Rerun when
model artifact revision change; LiteRT-LM version change; fixture or quality-gate change; browser/WebGPU behavior change; Input modality preprocessing or modality-specific quality gates
Per-case quality & failure details
[
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 1363.4650000333786,
      "version": "0.19.0",
      "artifactRevision": "ec14a83775e46685aad47bff6bbb647f6c3d1c4f",
      "loadMs": 130876.9849999547,
      "inferenceMs": 1363.4650000333786,
      "totalMs": 132240.44999998808,
      "charactersPerSecond": 3.6671275022663554,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 694.3349999785423,
      "version": "0.19.0",
      "artifactRevision": "ec14a83775e46685aad47bff6bbb647f6c3d1c4f",
      "loadMs": 130876.9849999547,
      "inferenceMs": 694.3349999785423,
      "totalMs": 131571.31999993324,
      "charactersPerSecond": 43.20680939449562,
      "grade": "fast",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls.",
      "imagesPerSecond": 1.4402269798165208
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 643.8700000047684,
      "version": "0.19.0",
      "artifactRevision": "ec14a83775e46685aad47bff6bbb647f6c3d1c4f",
      "loadMs": 130876.9849999547,
      "inferenceMs": 643.8700000047684,
      "totalMs": 131520.85499995947,
      "charactersPerSecond": 20.190411107682802,
      "grade": "fast",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls.",
      "imagesPerSecond": 1.553108546744831
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 515.3700000047684,
      "version": "0.19.0",
      "artifactRevision": "ec14a83775e46685aad47bff6bbb647f6c3d1c4f",
      "loadMs": 130876.9849999547,
      "inferenceMs": 515.3700000047684,
      "totalMs": 131392.35499995947,
      "charactersPerSecond": 1.940353532395653,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls.",
      "imagesPerSecond": 1.940353532395653
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 34615.75499999523,
      "version": "0.19.0",
      "artifactRevision": "ec14a83775e46685aad47bff6bbb647f6c3d1c4f",
      "loadMs": 130876.9849999547,
      "inferenceMs": 34615.75499999523,
      "totalMs": 165492.73999994993,
      "charactersPerSecond": 3.062189456795456,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 17063.889999985695,
      "version": "0.19.0",
      "artifactRevision": "ec14a83775e46685aad47bff6bbb647f6c3d1c4f",
      "loadMs": 130876.9849999547,
      "inferenceMs": 17063.889999985695,
      "totalMs": 147940.8749999404,
      "charactersPerSecond": 2.5785445171081673,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 17219.155000030994,
      "version": "0.19.0",
      "artifactRevision": "ec14a83775e46685aad47bff6bbb647f6c3d1c4f",
      "loadMs": 130876.9849999547,
      "inferenceMs": 17219.155000030994,
      "totalMs": 148096.1399999857,
      "charactersPerSecond": 0.5226737316659151,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 2191.889999985695,
      "version": "0.19.0",
      "artifactRevision": "ec14a83775e46685aad47bff6bbb647f6c3d1c4f",
      "loadMs": 130876.9849999547,
      "inferenceMs": 2191.889999985695,
      "totalMs": 133068.8749999404,
      "charactersPerSecond": 3.6498181934550598,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 1730.5450000166893,
      "version": "0.19.0",
      "artifactRevision": "ec14a83775e46685aad47bff6bbb647f6c3d1c4f",
      "loadMs": 130876.9849999547,
      "inferenceMs": 1730.5450000166893,
      "totalMs": 132607.5299999714,
      "charactersPerSecond": 4.622821134337939,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  }
]

Gemma 4 E4B Instruct — Text, Image & Audio

18 configurations tested
Passed18
Failed0
TestedNov 5, 2026
RuntimeTransformers.js 4.3.2

Quality criteria

{ "version": "multimodal-v1", "scope": "Text, image recognition/OCR/counting, speech and silence fixtures; per-fixture quality and genuine streaming parity." }

PrecisionDeviceResult
Q4F16GPU · WebGPUAcceleratedPassedFast
Q4GPU · WebGPUAcceleratedPassedFast
Q4F16GPU · WebGPUAcceleratedPassedModerate
Q4GPU · WebGPUAcceleratedPassedModerate
Q4F16GPU · WebGPUAcceleratedPassedModerate
Q4F16GPU · WebGPUAcceleratedPassedModerate
Q4F16GPU · WebGPUAcceleratedPassedModerate
Q4GPU · WebGPUAcceleratedPassedModerate
Q4F16GPU · WebGPUAcceleratedPassedModerate
Q4GPU · WebGPUAcceleratedPassedModerate
Q4GPU · WebGPUAcceleratedPassedModerate
Q4F16GPU · WebGPUAcceleratedPassedSlow
Q4GPU · WebGPUAcceleratedPassedSlow
Q4GPU · WebGPUAcceleratedPassedSlow
Q4F16GPU · WebGPUAcceleratedPassedSlow
Q4F16GPU · WebGPUAcceleratedPassedSlow
Q4GPU · WebGPUAcceleratedPassedVery slow
Q4GPU · WebGPUAcceleratedPassedVery slow
Test setup & sample details
Quality gate
{ "version": "multimodal-v1", "scope": "Text, image recognition/OCR/counting, speech and silence fixtures; per-fixture quality and genuine streaming parity." }
Tested
2026-11-05T17:29:37.138Z
Environment
Codex in-app Chromium on macOS; LiteRT-LM JS 0.17.0; WebGPU
Runtime
Transformers.js 4.3.2
Artifact revision
2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2
Fixture
{
  "file": "model-samples/multimodal/gemma-4-e4b-it-litert-lm-text-fixture.json",
  "url": "https://assets.localai.tools/samples/v1/files/b44d7be442f7c7f5c4948ad305192078d7a23a7e0210e1afe93c67461f6f066a/gemma-4-e4b-it-litert-lm-text-fixture.json",
  "size": 431,
  "sha1": "e617fd367ad31a5b536e4685e09de1f92f5e7e01",
  "sha256": "b44d7be442f7c7f5c4948ad305192078d7a23a7e0210e1afe93c67461f6f066a"
}
Rerun when
model artifact revision change; LiteRT-LM version change; fixture or quality-gate change; browser/WebGPU behavior change; Input modality preprocessing or modality-specific quality gates
Per-case quality & failure details
[
  {
    "precision": "q4f16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 969.6949999928474,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 3229.0099999904633,
      "inferenceMs": 969.6949999928474,
      "totalMs": 4198.704999983311,
      "charactersPerSecond": 5.15626047369212,
      "grade": "good",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4f16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 1441.1449999809265,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 3229.0099999904633,
      "inferenceMs": 1441.1449999809265,
      "totalMs": 4670.15499997139,
      "charactersPerSecond": 4.163356220282768,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls.",
      "imagesPerSecond": 0.6938927033804614
    }
  },
  {
    "precision": "q4f16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 1024.6949999928474,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 3229.0099999904633,
      "inferenceMs": 1024.6949999928474,
      "totalMs": 4253.704999983311,
      "charactersPerSecond": 12.68670189675049,
      "grade": "good",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls.",
      "imagesPerSecond": 0.9759001459038837
    }
  },
  {
    "precision": "q4f16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 888.2350000143051,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 3229.0099999904633,
      "inferenceMs": 888.2350000143051,
      "totalMs": 4117.245000004768,
      "charactersPerSecond": 1.1258281873421954,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls.",
      "imagesPerSecond": 1.1258281873421954
    }
  },
  {
    "precision": "q4f16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 2129.395000040531,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 3229.0099999904633,
      "inferenceMs": 2129.395000040531,
      "totalMs": 5358.405000030994,
      "charactersPerSecond": 50.249014390455194,
      "grade": "fast",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4f16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 623.0849999785423,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 3229.0099999904633,
      "inferenceMs": 623.0849999785423,
      "totalMs": 3852.0949999690056,
      "charactersPerSecond": 70.61636855567903,
      "grade": "fast",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4f16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 352.6650000214577,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 3229.0099999904633,
      "inferenceMs": 352.6650000214577,
      "totalMs": 3581.675000011921,
      "charactersPerSecond": 25.519969374484003,
      "grade": "fast",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4f16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 3828.9950000047684,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 3229.0099999904633,
      "inferenceMs": 3828.9950000047684,
      "totalMs": 7058.004999995232,
      "charactersPerSecond": 2.0893210881680537,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4f16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 3790.7799999713898,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 3229.0099999904633,
      "inferenceMs": 3790.7799999713898,
      "totalMs": 7019.789999961853,
      "charactersPerSecond": 2.110383614997541,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 1100.3199999928474,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 250119.93999999762,
      "inferenceMs": 1100.3199999928474,
      "totalMs": 251220.25999999046,
      "charactersPerSecond": 4.544132615995803,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 2278.560000002384,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 250119.93999999762,
      "inferenceMs": 2278.560000002384,
      "totalMs": 252398.5,
      "charactersPerSecond": 2.6332420476062612,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls.",
      "imagesPerSecond": 0.4388736746010435
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 1857.175000011921,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 250119.93999999762,
      "inferenceMs": 1857.175000011921,
      "totalMs": 251977.11500000954,
      "charactersPerSecond": 6.999878848205772,
      "grade": "good",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls.",
      "imagesPerSecond": 0.5384522190927518
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 1664.8450000286102,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 250119.93999999762,
      "inferenceMs": 1664.8450000286102,
      "totalMs": 251784.78500002623,
      "charactersPerSecond": 0.6006565175633859,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls.",
      "imagesPerSecond": 0.6006565175633858
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 2537.1299999952316,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 250119.93999999762,
      "inferenceMs": 2537.1299999952316,
      "totalMs": 252657.06999999285,
      "charactersPerSecond": 42.17363714125847,
      "grade": "fast",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 717.1750000119209,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 250119.93999999762,
      "inferenceMs": 717.1750000119209,
      "totalMs": 250837.11500000954,
      "charactersPerSecond": 61.35183183918657,
      "grade": "fast",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 477.72000002861023,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 250119.93999999762,
      "inferenceMs": 477.72000002861023,
      "totalMs": 250597.66000002623,
      "charactersPerSecond": 18.83948756480993,
      "grade": "good",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 6716.550000011921,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 250119.93999999762,
      "inferenceMs": 6716.550000011921,
      "totalMs": 256836.49000000954,
      "charactersPerSecond": 1.1910876863844981,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "Fixed authored/retained fixture; not broad task-quality certification."
    },
    "runtime": {
      "durationMs": 6832.009999990463,
      "version": "4.3.2",
      "artifactRevision": "2b6d2e26a962f70a33da8f7cd81e1afdd8d9cfb2",
      "loadMs": 250119.93999999762,
      "inferenceMs": 6832.009999990463,
      "totalMs": 256951.94999998808,
      "charactersPerSecond": 1.1709584734230727,
      "grade": "slow",
      "totalTimingMethod": "Sum of measured initialization and this inference; excludes preceding fixture calls."
    }
  }
]

Something not quite right?

Tell us what happened with Gemma 4. 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.