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.

GLiClass 4 Desktop

Rank custom labels against text using an optional instruction.

Desktop inferencePrivate · Runs on your computer
About this model

Rank custom labels against text using an optional instruction.

License
Apache-2.0
Model released
Unknown
Source repository created
Oct 16, 2026
Model weights · FP32
881.9 MB

Best for

  • Instruction-guided classification, routing and label ranking.

Keep in mind

  • Checked on bounded English classification fixtures. Scores depend on candidate wording and are not verified probabilities.
  • The combined instruction, labels and content is limited to 2,048 tokens.
  • Review sensitive decisions. The runtime returns complete label scores after inference.

Processed on your deviceYour input is processed locally in the desktop app. Model and runtime files download from external hosts.

Target platforms · verification required

  • macos · aarch64, x86_64
  • windows · x86_64

Recommended configuration

FP32 · cpu (verified platform only)

Sources & technical details
Original model
knowledgator/gliclass-v4-edge
Release verification
Unknown
Release evidence
Official collection names this exact checkpoint. An exact weight publication date was not verified.
Source updated
Oct 24, 2026
Native model files
knowledgator/gliclass-v4-edge
Files updated
Oct 24, 2026
Artifact revision
885af1af8a07edf4ed34df83ca0d995ed8201283

Desktop app

Sample result

GLiClass 4 Edge 0.4B Desktop

Run GLiClass 4 Desktop on your computer

Download the desktop app to load this model, provide input and view results locally.

Public desktop downloads are coming soon.

Device requirements

macOS Apple SiliconCPU / FP32Tested8.0 GiBRAM estimate
macOS Apple SiliconMPS / BF16Tested8.0 GiBRAM estimate
macOS Apple SiliconMPS / FP16Tested8.0 GiBRAM estimate
macOS Apple SiliconMPS / FP32Tested8.0 GiBRAM estimate
If your device falls short

MemoryAdvisory only: close other apps, reduce image/tile/token/batch settings or choose a smaller verified model/precision. CPU can need more RAM than GPU.

GPUIf the selected GPU runtime fails, existing CPU fallback may recover when enabled, but CPU has its own memory/runtime needs.

PlatformUse a compatible OS/architecture/runtime or another model. These estimates do not add hardware detection or block runs.

Benchmark evidence

Saved desktop measurements apply to the recorded hardware and runtime. Fallback cases report the device that actually ran; they do not prove GPU support on the requested device.

GLiClass 4 Edge 0.4B Desktop

4 configurations tested
Passed4
Failed0
TestedNov 14, 2026
Runtimetransformers 5.22.1

Quality criteria

Expected top label, relevant labels in top three, complete finite scores and less than 1.5 seconds per fixture.

PrecisionDeviceResultPlatformInference / load
BF16mpsAcceleratedPassedNot measuredmacos · aarch64205 ms / 2464 ms
FP16mpsAcceleratedPassedNot measuredmacos · aarch64205 ms / 2382 ms
FP32mpsAcceleratedPassedNot measuredmacos · aarch64203 ms / 2480 ms
FP32cpuPassedNot measuredmacos · aarch6498 ms / 4240 ms
Test setup & sample details
Quality gate
Expected top label, relevant labels in top three, complete finite scores and less than 1.5 seconds per fixture.
Tested
2026-11-14T13:18:43.557094+00:00
Environment
Apple M3 Max macOS 27.2. Three repetitions per fixture with independently initialized backends.
Runtime
transformers 5.22.1
Artifact revision
885af1af8a07edf4ed34df83ca0d995ed8201283
Fixture
{
  "file": "model-samples/gliclass-v4-product.json",
  "size": 279,
  "sha1": "aeb6baa836a422081758d97527bb1e922270d633",
  "sha256": "ae5cc91185924e774a5de326dfb98d21203b41991a912c7dc6a61fca61f62027",
  "url": "https://assets.localai.tools/samples/v1/files/ae5cc91185924e774a5de326dfb98d21203b41991a912c7dc6a61fca61f62027/gliclass-v4-product.json",
  "durationSeconds": 0
}
Rerun when
artifact or runtime revision; tokenizer or prompt formatting; fixture or threshold
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "matches": 1,
      "topLabel": "billing support",
      "grade": "ranking-pass"
    },
    "runtime": {
      "inferenceMs": 97.81041700625792,
      "durationMs": 97.81041700625792,
      "itemsPerSecond": 10.223859897622356,
      "peakResidentMemoryBytes": 3202891776,
      "outputDevice": "cpu",
      "loadMs": 4239.581832953263,
      "totalMs": 4337.392249959521,
      "grade": "interactive",
      "engine": "transformers",
      "version": "5.22.1",
      "device": "cpu",
      "precision": "fp32",
      "fallback": false,
      "artifactRevision": "885af1af8a07edf4ed34df83ca0d995ed8201283",
      "torchVersion": "2.10.0",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": [
        "cpu/torch.float32"
      ],
      "cpuThreads": 4,
      "requestedDevice": "cpu",
      "requestedPrecision": "fp32"
    }
  },
  {
    "precision": "bf16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "matches": 1,
      "topLabel": "billing support",
      "grade": "ranking-pass"
    },
    "runtime": {
      "inferenceMs": 204.7680409741588,
      "durationMs": 204.7680409741588,
      "itemsPerSecond": 4.8835745814758145,
      "peakResidentMemoryBytes": 675004416,
      "outputDevice": "mps",
      "loadMs": 2464.3356659798883,
      "totalMs": 2669.103706954047,
      "grade": "interactive",
      "engine": "transformers",
      "version": "5.22.1",
      "device": "mps",
      "precision": "bf16",
      "fallback": false,
      "artifactRevision": "885af1af8a07edf4ed34df83ca0d995ed8201283",
      "torchVersion": "2.10.0",
      "provider": "PyTorch Metal",
      "acceleration": "partial",
      "placement": [
        "mps:0/torch.bfloat16"
      ],
      "cpuThreads": 4,
      "requestedDevice": "mps",
      "requestedPrecision": "bf16"
    }
  },
  {
    "precision": "fp16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "matches": 1,
      "topLabel": "billing support",
      "grade": "ranking-pass"
    },
    "runtime": {
      "inferenceMs": 205.4120830143802,
      "durationMs": 205.4120830143802,
      "itemsPerSecond": 4.868262788270315,
      "peakResidentMemoryBytes": 2103115776,
      "outputDevice": "mps",
      "loadMs": 2381.668125046417,
      "totalMs": 2587.0802080607973,
      "grade": "interactive",
      "engine": "transformers",
      "version": "5.22.1",
      "device": "mps",
      "precision": "fp16",
      "fallback": false,
      "artifactRevision": "885af1af8a07edf4ed34df83ca0d995ed8201283",
      "torchVersion": "2.10.0",
      "provider": "PyTorch Metal",
      "acceleration": "partial",
      "placement": [
        "mps:0/torch.float16"
      ],
      "cpuThreads": 4,
      "requestedDevice": "mps",
      "requestedPrecision": "fp16"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "matches": 1,
      "topLabel": "billing support",
      "grade": "ranking-pass"
    },
    "runtime": {
      "inferenceMs": 202.75029097683728,
      "durationMs": 202.75029097683728,
      "itemsPerSecond": 4.932175412336363,
      "peakResidentMemoryBytes": 2989899776,
      "outputDevice": "mps",
      "loadMs": 2480.079040979035,
      "totalMs": 2682.8293319558725,
      "grade": "interactive",
      "engine": "transformers",
      "version": "5.22.1",
      "device": "mps",
      "precision": "fp32",
      "fallback": false,
      "artifactRevision": "885af1af8a07edf4ed34df83ca0d995ed8201283",
      "torchVersion": "2.10.0",
      "provider": "PyTorch Metal",
      "acceleration": "partial",
      "placement": [
        "mps:0/torch.float32"
      ],
      "cpuThreads": 4,
      "requestedDevice": "mps",
      "requestedPrecision": "fp32"
    }
  }
]
Release validation still required
  • Model/runtime backup and dependent outage checks deferred.
  • Windows benchmarking deferred.

Something not quite right?

Tell us what happened with GLiClass 4 Desktop. 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.