Loading comparison…
HEAD TO HEAD
HIT-TMG (Lychee Team): jevembed and Rizzo AI Academy: rizzo-flow, compared on what they decide, where they run, what they cost and what their publishers report.
| Property | hit-tmg/jevembed | rizzo-ai-academy/rizzo-flow |
|---|---|---|
| Summary | An embedding model turned decision model. Qwen3-Embedding-4B with a merged LoRA, driven by the open JevEmbed toolkit, which scores Choice, Score and Noul questions from embedding similarity and returns a distribution without generating text. | A local, Jev-compatible decision model from Rizzo AI Academy. XHToken's Spark-X2.5-4B with a merged typed-decisions LoRA, run on llama.cpp; yes/no, choice and score questions share one prefill of the state and are read from the answer-letter logits. No text is generated. |
| Decides | choice, score, noul, classify | choice, score, noul, classify, route |
| Architecture | jevembed | rizzo-flow |
| Fine-tuned from | qwen/qwen3-embedding-4b | xhtoken/spark-x2.5-4b |
| License | apache-2.0 | apache-2.0 |
| Availability | Open weights | Open weights |
| Hosted by | — | — |
| Input price | — | — |
| Decision accuracy | 85.9% | 64.8% |
| Calibration error | — | 0.112 |
| Valid action rate | — | — |
| Median latency | — | 195 ms |
| p95 latency | — | — |
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.
jevembed is from HIT-TMG (Lychee Team) and rizzo-flow from Rizzo AI Academy. Both have open weights you can download and run. Both answer choice, score, noul and classify questions. Only rizzo-flow answers route.
They report on different suites — jevembed 85.9% on JevEmbed-Data test split (64,110 hard-label questions), final LoRA checkpoint before merging, rizzo-flow 64.8% on LocalLLaMA/typed-decisions test split (400 cases, 2,000 decisions), Q8_0 — so the numbers do not rank them. Test both on your own labelled examples.
jevembed: Free (open weights). rizzo-flow: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Yes, both: systemone pull hit-tmg/jevembed and systemone pull rizzo-ai-academy/rizzo-flow download the weights.
| Evaluation suite | JevEmbed-Data test split (64,110 hard-label questions), final LoRA checkpoint before merging | LocalLLaMA/typed-decisions test split (400 cases, 2,000 decisions), Q8_0 |
|---|
| Latest version | 2026.09 | 2026.09 |
|---|
| Variants | NOTICE, 1_Pooling, LICENSE, lora | lora, LICENSE |
|---|
| Size of latest version | 15.1 GB | 22.0 GB |
|---|
| Files | 21 | 17 |
|---|
| Downloads | 0 | 0 |
|---|
| Stars | 0 | 0 |
|---|
| Tags | system-one, embedding, qwen, lora, sentence-transformers, 4b | system-one, spark, lora, gguf, llama-cpp, 4b |
|---|
| Updated | Sep 30, 2026 | Sep 30, 2026 |
|---|