HEAD TO HEAD
Dimitris Papadopoulos: biodecision and Contrastive-LM: clm, compared on what they decide, where they run, what they cost and what their publishers report.
| Property | lighteternal/biodecision | contrastive-lm/clm |
|---|---|---|
| Summary | A 4B biomedical decision model by Dimitris Papadopoulos: Qwen3.5-4B tuned with Together AI's Tev1 recipe on 1.75M pharma, clinical-trial and literature decisions. One forward pass gives a temperature-scaled probability per option. Research use only. | Contrastive Language Models score a state against a set of candidate actions with a contrastive objective. Two projection heads on frozen Qwen3-8B embeddings, trained with InfoNCE; clm-serve maps Choice, Score and Noul onto candidate ranking. |
| Decides | choice, score, noul, classify | choice, score, noul, rank, classify, route |
| Architecture | biodecision | clm |
| Fine-tuned from | qwen/qwen3.5-4b | qwen/qwen3-8b |
| License | Research use only (several training sources are non-commercial); not validated for patient care | apache-2.0 |
| Availability | Open weights | Open weights |
| Hosted by | — | — |
| Input price | — | — |
| Decision accuracy | 75.3% | — |
| Calibration error | — | — |
| Valid action rate | — | — |
| Median latency | 39 ms | — |
| p95 latency | — | — |
| Evaluation suite | BioDecision v2 held-out test set (63,432 overlap-screened decisions from public biomedical benchmarks; the maker's own split) | — |
| Latest version | 2.0.0 | 2026.09.25 |
| Variants | LICENSE | LICENSE |
| Size of latest version | 8.7 GB | 72.1 MB |
| Files | 15 | 3 |
| Downloads | 0 | 45 |
| Stars | 0 | 0 |
| Tags | system-one, qwen, biomedical, clinical-trials, jev-inspired, research-only, 4b | system-one, contrastive, qwen, ranking, 8b |
| Updated | Oct 7, 2026 | Nov 21, 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.
biodecision is from Dimitris Papadopoulos and clm from Contrastive-LM. Both have open weights you can download and run. Both answer choice, score, noul and classify questions. Only clm answers rank and route. biodecision is the smaller model, at 4.0B parameters to 8.0B. biodecision is licensed other; clm, apache-2.0.
Only biodecision publishes an accuracy figure (75.3% on BioDecision v2 held-out test set (63,432 overlap-screened decisions from public biomedical benchmarks; the maker's own split)); clm does not, so there is no comparison to make without your own test.
biodecision: Free (open weights). clm: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Yes, both: systemone pull lighteternal/biodecision and systemone pull contrastive-lm/clm download the weights.