HEAD TO HEAD
Avishek Biswas: bev-decider and Yoav Pinto: nativ-he-decision, compared on what they decide, where they run, what they cost and what their publishers report.
| Property | avishek-biswas/bev-decider | yoav-pinto/nativ-he-decision |
|---|---|---|
| Summary | A 0.4B decision model made from the first 20 layers of Qwen3-0.6B plus a small attention head that compares options. It answers choice, noul and score questions in one pass, and choice order cannot change its answer by construction. Non-commercial licence. | A 364M Hebrew decision encoder on NeoDictaBERT: give it a state, a question and 2 to 7 free-text options and it returns a probability for each option in one forward pass, about 70 ms on a laptop CPU. Hebrew only. |
| Decides | choice, score, noul | choice, classify, route |
| Architecture | bev-decider | nativ |
| Fine-tuned from | qwen/qwen3-0.6b | dicta-il/neodictabert |
| License | cc-by-nc-4.0 | cc-by-4.0 |
| Availability | Open weights | Open weights |
| Hosted by | — | — |
| Input price | — | — |
| Decision accuracy | 74.7% | 96.2% |
| Calibration error | — | 0.010 |
| Valid action rate | — | — |
| Median latency | — | — |
| p95 latency | — | — |
| Evaluation suite | avbiswas/bev-decision test split (5,000 held-out questions over 2,617 states), the maker's own | nativ-bench user-intent task, 4 options (MASSIVE Hebrew, 2,973 items; the maker's own suite) |
| Latest version | 2026.09 | 2026.09 |
| Variants | LICENSE-Qwen | — |
| Size of latest version | 940.1 MB | 697.3 MB |
| Files | 8 | 5 |
| Downloads | 0 | 0 |
| Stars | 0 | 0 |
| Tags | system-one, qwen, non-commercial, 478m | system-one, hebrew, encoder, cpu, 364m |
| Updated | Oct 7, 2026 | Oct 7, 2026 |
Figures are from each model’s manifest; accuracy and latency are what the publishers report, on their own suites and hardware. Add a third model.
bev-decider is from Avishek Biswas and nativ-he-decision from Yoav Pinto. Both have open weights you can download and run. Both answer choice questions. Only bev-decider answers score and noul. Only nativ-he-decision answers classify and route. bev-decider reads up to 2K tokens of state, against 1K tokens for nativ-he-decision. nativ-he-decision is the smaller model, at 364M parameters to 478M. bev-decider is licensed cc-by-nc-4.0; nativ-he-decision, cc-by-4.0.
They report on different suites — bev-decider 74.7% on avbiswas/bev-decision test split (5,000 held-out questions over 2,617 states), the maker's own, nativ-he-decision 96.2% on nativ-bench user-intent task, 4 options (MASSIVE Hebrew, 2,973 items; the maker's own suite) — so the numbers do not rank them. Test both on your own labelled examples.
bev-decider: Free (open weights). nativ-he-decision: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Yes, both: systemone pull avishek-biswas/bev-decider and systemone pull yoav-pinto/nativ-he-decision download the weights.