GLiNER 2.5 small
FastinoOpen weights
The smallest GLiNER2 checkpoint.
- typed-decisions
- —
- Metatext suite
- 0.574
- ECE untuned
- 0.172
- p50 per case
- 132 ms
not run
4/7 over baseline
four CPU cores
Specifications
- Checkpoint
- fastino/gliner2.5-small-v1
- Publisher
- Fastino
- Mechanism
- Schema-conditioned encoder (extraction + classification)
- Backbone
- DeBERTa-v3 (boundary pairing)
- Parameters
- 74M
- License
- Apache 2.0
- Runs as
- open weights, self-hosted
- Library
- gliner2
- Primitives
- Choice, Score, Noul, Extraction
- Pricing
- none — your compute
Metatext decisionsrank 9 of 21
Full leaderboard →Per-question accuracy — default payload, serial mode, 120 judgments
| Task | Question | Type | Accuracy | Baseline | Beats? |
|---|---|---|---|---|---|
| agent_traces | outcome | choice | 0.429 | 0.357 | ✅ |
| agent_traces | needs_human | noul | 0.714 | 0.714 | ❌ |
| customer_service | department | choice | 0.650 | 0.400 | ✅ |
| customer_service | refund_requested | noul | 0.650 | 0.700 | ❌ |
| customer_service | frustration | score | 0.750 | 0.500 | ✅ |
| security_incidents | severity | choice | 0.438 | 0.312 | ✅ |
| security_incidents | actively_exploited | noul | 0.500 | 0.688 | ❌ |
Frequently asked questions
How does GLiNER 2.5 small compare to Jev?
On the Metatext suite GLiNER 2.5 small scores 0.574 against Jev's 0.863.
Is GLiNER 2.5 small open source?
Yes. fastino/gliner2.5-small-v1 is released under Apache 2.0 and runs on your own hardware with the gliner2 library.
Can GLiNER 2.5 small extract entities?
Yes. GLiNER 2.5 small has a span head and scores 0.812 micro-F1 on the eight-case extraction task.