SmolLM2-1.7B-Instruct (q4f16_1, WebLLM)

models.skillsafe.ai/smollm2-1.7b-instruct-q4f16@84f57f8580a9/

Vetted New WebGPU

Text generation model for web-llm (q4f16_1). Runs on WebGPU — 924 MB downloaded once from models.skillsafe.ai, then cached for every SkillSafe app that uses it. Inference happens on your device; nothing you enter is uploaded to load it, and it never costs a credit. Weights from Hugging Face · skillsafe-ai/smollm2-1.7b-instruct-q4f16, pinned at a0721139dff8 — also loadable straight from Hugging Face outside SkillSafe (how).

text generation web-llm WebGPU q4f16_1 Apache-2.0
Use it in an app Browse all models Request a variant
Share

Details

Catalogue idsmollm2-1.7b-instruct-q4f16@84f57f8580a9
Tasktext generation
Runtimeweb-llm ≥ 0.2.80
DeviceWebGPU
Variantq4f16_1
Download924 MB · 38 files
LicenceApache-2.0 · notice
SourceHugging Face · skillsafe-ai/smollm2-1.7b-instruct-q4f16
Pinned ata0721139dff8ab5a80c5c57e0df0f3871a2db426
ApprovedSep 22, 2026
Statusactive — every file is a live vetted hash
API/v1/browser-models/smollm2-1.7b-instruct-q4f16

Files

Each file is served at an immutable URL; the canonical form is the SHA-256 itself. Tokenizer and config JSON are not here by design — they ship in your app bundle.

PathFormatSizeSHA-256
params_shard_0.bin bin 48.0 MB 669337a7863d…864b50
params_shard_1.bin bin 31.0 MB 3b06106f1352…51a056
params_shard_10.bin bin 27.0 MB dc6a1862de63…ad5bc7
params_shard_11.bin bin 16.0 MB 374e196842dc…2cdd90
params_shard_12.bin bin 29.0 MB 3baf0c943fcf…c79c9c
params_shard_13.bin bin 27.0 MB 12b9152eb50a…3a56cf
params_shard_14.bin bin 16.0 MB bbdd552b53ff…4d99d0
params_shard_15.bin bin 29.0 MB bd1746861c89…abf132
params_shard_16.bin bin 27.0 MB d48924646b7c…1ec805
params_shard_17.bin bin 16.0 MB 490869285484…420c82
params_shard_18.bin bin 29.0 MB 91bea1e3348d…8df35f
params_shard_19.bin bin 27.0 MB c5abaa24e0c7…af6352
params_shard_2.bin bin 16.0 MB 0a3d8fb171aa…1c0dac
params_shard_20.bin bin 16.0 MB 3863eee9b385…670458
params_shard_21.bin bin 29.0 MB 269f85ea4c93…682d63
params_shard_22.bin bin 27.0 MB 946b4ec4c31f…853d47
params_shard_23.bin bin 16.0 MB f73d79db598f…60af47
params_shard_24.bin bin 29.0 MB 8d71924c6d5c…761eff
params_shard_25.bin bin 27.0 MB a59e7b9cfccc…115fde
params_shard_26.bin bin 16.0 MB ada169dcb95b…f941ba
params_shard_27.bin bin 29.0 MB f11613319a9a…ae9240
params_shard_28.bin bin 27.0 MB d9da814961c6…770133
params_shard_29.bin bin 16.0 MB 02424ed87af5…f52127
params_shard_3.bin bin 31.0 MB 1db95b8c0765…090b0a
params_shard_30.bin bin 29.0 MB 3f6a2c8aa692…b60907
params_shard_31.bin bin 27.0 MB 7a4373d8e1b7…db1dbf
params_shard_32.bin bin 16.0 MB 557133b873d9…93184c
params_shard_33.bin bin 29.0 MB 6c85b99eb0d1…2c6c1a
params_shard_34.bin bin 27.0 MB 57d8bff6cf65…680c7b
params_shard_35.bin bin 16.0 MB 1a18e8d1c838…165a9f
params_shard_36.bin bin 29.0 MB 2782bf188454…24c27f
params_shard_4.bin bin 27.0 MB 990d61768f72…80ed15
params_shard_5.bin bin 16.0 MB 70b337f4a1e3…3dbfad
params_shard_6.bin bin 29.0 MB 5a0816a9e5d1…b0e3ad
params_shard_7.bin bin 27.0 MB be743b234e9a…01fad0
params_shard_8.bin bin 16.0 MB 8e054bc56e90…71c19d
params_shard_9.bin bin 29.0 MB e3866cca4f51…63f7cf
lib/SmolLM2-1.7B-Instruct-q4f16_1-ctx4k_cs1k-webgpu.wasm wasm 5.6 MB 2a7d1f62079c…454039
Use it in an app declaration · SDK loader · web-llm · URLs · Hugging Face — generated from this entry

Add to the body of POST /v1/apps/{slug}/releases (or a release session). An unknown or withdrawn model is refused with a 400 naming it; the app page then shows "downloads 924 MB · runs on your device" and /models.txt carries the attribution.

{
  "models": [
    {
      "id": "smollm2-1.7b-instruct-q4f16",
      "revision": "84f57f8580a9"
    }
  ]
}

Evaluation

shard sizes + md5 vs ndarray-cache.json; implied parameter count vs the checkpoint; parameter table vs model_lib embedded metadata — MLC's published weights, checked against the WebGPU library. Evaluated Sep 22, 2026.

Library fit

params_matched: 243 · symbolic_dims_resolved: ["model.embed_tokens.q_scale","model.embed_tokens.q_weight"] · model_type: llama · quantization: q4f16_1 · effective: {"context_window_size":4096,"prefill_chunk_size":1024}

Toolchain: python 3.12.13 · platform Darwin 25.6.0 arm64 · torch 2.10.0 · onnx 1.23.0 · onnxruntime 1.30.0. Recipe models/recipes/smollm2-1.7b-instruct-q4f16.yaml (64ddf52e7506). Full manifest.json

Licence & attribution

Apache-2.0 · licence text · notice

SmolLM2-1.7B-Instruct: Hugging Face TB, Apache License 2.0. https://huggingface.co/HuggingFaceTB/SmolLM2-1.7B-Instruct — q4f16_1 weights published by MLC (https://huggingface.co/mlc-ai/SmolLM2-1.7B-Instruct-q4f16_1-MLC).

Apps that declare this model get this text in their generated /models.txt, so a licence that requires a notice always carries one.

Revisions

RevisionVariantSizeApprovedStatusApps
smollm2-1.7b-instruct-q4f16@84f57f8580a9 q4f16_1 924 MB Sep 22, 2026 active 0
smollm2-1.7b-instruct-q4f16@84f57f85 q4f16_1 918 MB Sep 22, 2026 withdrawn 0

Every file here was approved by exact SHA-256 after a structural audit of the graph, fetched from a content-pinned source, and is served credential-free at an immutable URL. The SDK re-verifies the hash on your device before it caches or returns anything. Missing a variant? Request it — or read how the registry works.