Loading comparison…
HEAD TO HEAD
StartLux: startlux-decision and Nandakishor M: vega, compared on what they decide, where they run, what they cost and what their publishers report.
| Property | startlux/startlux-decision | nandakishorm/vega |
|---|---|---|
| Summary | StartLux's decision family in six sizes, 0.8B to a 35B-A3B mixture of experts. All questions in a request are answered in one forward pass, with a probability for every option, through a TypeSafe /v1/systemone-compatible server. Text, JSON or images, up to 256K tokens. | Nandakishor M's typed-decision model. A frozen Qwen3.5 backbone is read, never trained, and cut off mid-stack; a small trained physics engine settles each choice, score or yes/no question into calibrated probabilities, a conformal set and an abstain flag. 0.8B and 4B. |
| Decides | choice, score, noul | choice, score, noul, classify, route |
| Architecture | startlux-decision | vega |
| Fine-tuned from | — | qwen/qwen3.5-0.8b |
| License | cc-by-nc-4.0 | apache-2.0 |
| Availability | Open weights | Open weights |
| Hosted by | — | — |
| Input price | — | — |
| Decision accuracy | 88.3% | 55.6% |
| Calibration error | — | 0.151 |
| Valid action rate | — | — |
| Median latency | 26 ms | — |
| p95 latency | — | — |
| Evaluation suite | JevBench public set (231 items; 204 correct), maker's run | His 11 October 2026 benchmark, published with its results and its harness (a notebook in vegaml): macro accuracy over 43 public tasks for the 0.8B, 0.717 for the 4B; with JevBench (231 items) and DecisionBench (1,075 rows), 3,841 decisions in all, seed 20261011. On his own set of 2,050 choice decisions over six intent workflows (esci, clinc150, appliances, banking77, when2call, apibank), one choice question each, which he has not published with its question wording, he reports 0.763 (calibration error 0.026) for the 0.8B with its public_intents adapter, 0.389 without the adapter and 0.803 for the 4B with it. He publishes no accuracy for the typed_decisions adapters |
| Latest version | 2026.10 | 2026.10.12 |
| Variants | LICENSE, NOTICE, startlux_decision | adapters, backbone, 4b |
| Size of latest version | 8.7 GB | 10.5 GB |
| Files | 23 | 31 |
| Downloads | 3 | 108 |
| Stars | 0 | 0 |
| Tags | system-one, qwen, non-commercial, gguf, cuda-graphs, vision, mlx, 4.7b | system-one, qwen, frozen-backbone, calibrated, conformal, abstention, adapters, multimodal, long-context, 0.8b |
| Updated | Oct 7, 2026 | Oct 29, 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.
startlux-decision is from StartLux and vega from Nandakishor M. Both have open weights you can download and run. Both answer choice, score and noul questions. Only vega answers classify and route. startlux-decision reads up to 256K tokens of state, against 72K tokens for vega. vega is the smaller model, at 800M parameters to 4.7B. startlux-decision is licensed cc-by-nc-4.0; vega, apache-2.0.
They report on different suites — startlux-decision 88.3% on JevBench public set (231 items; 204 correct), maker's run, vega 55.6% on His 11 October 2026 benchmark, published with its results and its harness (a notebook in vegaml): macro accuracy over 43 public tasks for the 0.8B, 0.717 for the 4B; with JevBench (231 items) and DecisionBench (1,075 rows), 3,841 decisions in all, seed 20261011. On his own set of 2,050 choice decisions over six intent workflows (esci, clinc150, appliances, banking77, when2call, apibank), one choice question each, which he has not published with its question wording, he reports 0.763 (calibration error 0.026) for the 0.8B with its public_intents adapter, 0.389 without the adapter and 0.803 for the 4B with it. He publishes no accuracy for the typed_decisions adapters — so the numbers do not rank them. Test both on your own labelled examples.
startlux-decision: Free (open weights). vega: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Yes, both: systemone pull startlux/startlux-decision and systemone pull nandakishorm/vega download the weights.