Up to date
fastino Apache 2.0 out 2 months ago
287M parameters.
Verified 2026-10-26 07:27 UTC Rank #0 in Decision models · score 0 method v1 Sources 3 Downloads 30d 132.6k · #285 of 619 tracked
Picked by rule from 1 tracked piece: the lab's own post, the most shared, a field report, the latest.
GLiNER2.5-Decide is a 340M-parameter DeBERTa-v3-large encoder released under Apache-2.0 that runs natively on Macs via MLX Swift.
Hugging Face does not say who downloads a model or with what. These are the proxies: API providers listing it, quantized builds, fine-tunes/adapters/merges, and Spaces that load it.
No API provider serves it, so its downloads are people running it themselves: 7 quantized builds and 11 Spaces.
Top Spaces running it: hugging-apps/gliner2-5-multi-extractor ♥ 10 · fastino/gliner25-long-context ♥ 9 · fastino/gliner25-span-attributes ♥ 4 · fastino/gliner25-unlimited-span-length ♥ 2
Weights only, an estimate: leave room for the KV cache. Method
About 0.2 GB at INT4, 0.3 GB at FP8, 0.7 GB at BF16: weights only, so leave room for the KV cache.
Estimate params × bytes per param × 1.2 · how we estimate
Weights, every dated article, and repo commits on one rail.
Teal is the weights (2026-08-14), amber triangles are 1 dated article (each links to it), grey ticks are repo commits.
27 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.
From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.
Everything else, folded: the full tables, every source.
Labels are evidence, not verdicts: measured a third-party leaderboard row for this model; author-reported a number the model author published in the card or eval files; structural a deterministic signal from the name, config or chat template, or a decision-model estimate. Each chip links to its source. All use cases.
Listed in Decision models
| Card pipeline tag | schema-extraction |
|---|---|
| Library | transformers |
| Architectures | BoundaryExtractor |
| Model type | extractor |
| Params by dtype | F32: 287,355,159 |
| Size band | S |
| MoE | no |
| Languages | multilingual, en |
| License | Apache 2.0 (apache-2.0) |
| Hub snapshot | 2026-10-26 |
vLLM: not in registry · 4b1c04e76fe9ac3ca8b7e541dd423e7d5f38dc60SGLang: not in registry · 60d941293e3af27a12c6eb64e2c3e2e1a17f100f
vLLM registry (4b1c04e76fe9ac3ca8b7e541dd423e7d5f38dc60) · SGLang registry (60d941293e3af27a12c6eb64e2c3e2e1a17f100f) · Matched by architecture class name against each project's registry on main; 'supported' means the architecture is registered, not that this checkpoint was tested.
Weights-only memory for gliner2.5-multi-v1: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 0.7 GB · Edge (<4 GB) |
|---|---|
| FP8 (1 byte/param) | 0.3 GB · Edge (<4 GB) |
| INT4 (0.5 bytes/param) | 0.2 GB · Edge (<4 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.
| How it was made | own architecture (pre-existing encoder classifier) |
|---|---|
| Lineage bucket | O - original architecture or own base |
| Lineage evidence | Fastino own architecture; HF created 2026-08-14; prior art |
| DI 0.2.1 score | 4.26 (independent (DI 0.2.1 board)) |
| ECE (board) | 0.2538 |
| Median latency | 14 ms (DI board, 1x RTX PRO 6000, single process) |
| Hosted price (input) | self-hosted/M |
| Notes | Board raised a capacity limit ("lifted"). |
| Source | https://huggingface.co/fastino/gliner2.5-multi-v1 |
finetune: 19quantized: 7adapter: 1
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| fastino/GLiNER2.5-multi-Decide | finetune | — | 287M | S | apache-2.0 | fastino | 68 | 22k |
| newmindai/gliner2.5-kvkk-tr-v2 | finetune | — | 287M | S | apache-2.0 | newmindai | 0 | 2.8k |
| metrika-labs/ru-pii-ner-gliner2.5 | finetune | — | 287M | S | apache-2.0 | metrika-labs | 3 | 1.9k |
| furkanbekmezci/turkextract-turkish-information-extraction | finetune | — | 287M | S | apache-2.0 | furkanbekmezci | 0 | 711 |
| DanKau/gliner2.5-multi-v1-onnx | quantized | — | — | — | apache-2.0 | DanKau | 1 | 380 |
| FluidInference/gliner2-5-multi-coreml | quantized | — | — | — | apache-2.0 | FluidInference | 0 | 289 |
| Elafnawaf/gliner2-arabic-multitask | finetune | — | 287M | S | cc-by-sa-4.0 | Elafnawaf | 1 | 194 |
| matura-lol/GLiNER2.5-multi-Decide-finetune | finetune | — | 287M | S | apache-2.0 | matura-lol | 0 | 226 |
| litert-community/GLiNER2.5-Multi-LiteRT | finetune | — | — | — | apache-2.0 | litert-community | 0 | 150 |
| lucataco/gliner2.5-cua-grounder-macos-v1 | finetune | — | 287M | S | apache-2.0 | lucataco | 2 | 44 |
| helmo/GLiNER2.5-multi-Decide | finetune | — | 287M | S | apache-2.0 | helmo | 1 | 92 |
| Elafnawaf/gliner2-arabic-saudi | finetune | — | 287M | S | cc-by-sa-4.0 | Elafnawaf | 0 | 142 |
| lucataco/gliner2.5-cua-grounder-macos-v2 | finetune | — | 287M | S | apache-2.0 | lucataco | 1 | 41 |
| newmindai/gliner2.5-kvkk-tr-v1 | finetune | — | 287M | S | apache-2.0 | newmindai | 1 | 24 |
| Elafnawaf/gliner2-arabic-saudi-r4-l-contk-s42 | finetune | — | 287M | S | cc-by-sa-4.0 | Elafnawaf | 1 | 18 |
| AIbnuHibban/gliner2.5-indonesian | finetune | — | 287M | S | apache-2.0 | AIbnuHibban | 0 | 65 |
| hadimaster65555/gliner25-indonesian-sentiment | adapter | — | 287M | S | apache-2.0 | hadimaster65555 | 0 | 18 |
| Elafnawaf/gliner2-arabic-saudi-r3-f-full-s42 | finetune | — | 287M | S | cc-by-sa-4.0 | Elafnawaf | 0 | 17 |
| ranzucker/gliner2-5-multi-coreml | quantized | — | — | — | apache-2.0 | ranzucker | 0 | 13 |
| Elafnawaf/gliner2-arabic-saudi-r4-k-cont-s42 | finetune | — | 287M | S | cc-by-sa-4.0 | Elafnawaf | 0 | 8 |
| jugaadsrl/gliner2.5-multi-v1-onnx | quantized | — | — | — | apache-2.0 | jugaadsrl | 0 | 0 |
| nicolasembleton/gliner2.5-multi-v1-onnx | quantized | — | — | — | apache-2.0 | nicolasembleton | 0 | 0 |
| beshkenadze/gliner2.5-multi-onnx | quantized | — | — | — | apache-2.0 | beshkenadze | 0 | 0 |
| smdesai/GLiNER25-Multi-FP16-CoreAI | finetune | -FP16 | — | — | apache-2.0 | smdesai | 0 | 0 |
| smdesai/GLiNER25-Multi-FP16-CoreML | finetune | -FP16 | — | — | apache-2.0 | smdesai | 0 | 0 |
| IkerAzurmendi/bin-intent-gliner2 | finetune | — | — | — | apache-2.0 | IkerAzurmendi | 0 | 0 |
| Ashad001/vault-gliner2.5-multi-onnx | quantized | — | — | — | apache-2.0 | Ashad001 | 0 | 0 |
| Date | Commit | SHA |
|---|---|---|
| 2026-10-08 | Load this checkpoint as a Transformers schema-extraction model. (#11) | cf5593a |
| 2026-09-28 | Point the fine-tune banner to agent.fastino.ai | 2ca71aa |
| 2026-09-24 | Update the Fastino GLiNER skill | 12fc403 |
| 2026-09-24 | Update GLiNER2 deployment badge link (#10) | c43ffb1 |
| 2026-09-24 | Update GLiNER2.5 fine-tuning badge link (#9) | 6569aa2 |
| 2026-09-24 | Update Fastino fine-tuning link (#8) | 3619746 |
| 2026-09-24 | Update the Fastino GLiNER skill | 4921c00 |
| 2026-09-24 | Mention hosted fine-tuning via the Fastino API (#7) | 93de66e |
| 2026-09-24 | Update the Fastino GLiNER skill | 2333292 |
| 2026-09-24 | Add the Fastino GLiNER skill | 081094e |
| 2026-09-20 | update url (#6) | a221b77 |
| 2026-09-17 | Upload GLiNER-2.5-Multi-HF-Banner.png (#4) | 235cf92 |
| 2026-09-17 | update banner 2 (#5) | dc3e20a |
| 2026-08-22 | Update README.md | aaecfe4 |
| 2026-08-21 | Upload folder using huggingface_hub | 07dae4f |
| 2026-08-20 | Add example outputs to every usage snippet. | d75bf6a |
| 2026-08-20 | Add Fastino model card banner. | d782b12 |
| 2026-08-20 | Add GLiNER2.5 boundary architecture model card and banner. | a7b99bd |
| 2026-08-18 | Upload BoundaryExtractor | 2cc5320 |
| 2026-08-14 | Update README.md | 80239fa |
| 2026-08-14 | Update README.md | eddb3a2 |
| 2026-08-14 | Update README.md | 5c6f654 |
| 2026-08-14 | Upload BoundaryExtractor | d24d337 |
| 2026-08-14 | initial commit | 2c22f29 |
Ranked by source signals (HN, Lobsters, authority domains, official posts) and recency. Summaries are generated from a stored copy of the article text; every sentence cites a span of that text, and nothing is shown that failed the citation check.
arXiv links found in the model card and its tags. "Peer-reviewed" is shown only with evidence: a published venue in OpenAlex (measured) or the authors' own journal reference on arXiv (reported).
Hub snapshot: 2026-10-26 · HF API: 2026-10-26 07:25 UTC · Providers: 2026-10-26 07:26 UTC · Articles: 2026-10-26 07:27 UTC · Summary: 2026-10-07 15:18 UTC · Last verified 2026-10-26 07:27 UTC
Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.
