whatmodel.sh / decision / Jebadiah 27B v2.1 (Ollaya Q4_K_M)
Jebadiah 27B v2.1 (Ollaya Q4_K_M)
↗ share
+ contribute
Jebadiah 27B v2.1 (Ollaya Q4_K_M) is an open-weights decision model from frontier-infra, released October 2026. Licence: apache-2.0.
Catalogue identity notes
Ollaya native AINode prompt and per-type temperatures, one pass, Q4_K_M. This is not the repeated-payload/forward-reverse T1e recipe; its scores stay separate. Source GGUF jebadiah-27b-v2-1-Q4_K_M.gguf at ed71f41b8b59e8832c53982cd26e0879824e979c; weight sha256:f508eb3ad7b0bee0c9d7460288e0f56b764e5ad01d7bfba7d58c6a347cee4a3a; Ollaya manifest 978465d647d1de010ea24370f135b3b2552020208612163c5907abe92de465de. Published training-overlap subsets are excluded. Local memory is the measured RTX 5090 peak increase during the short five-question triage workload (16591 MiB), rounded upward to 0.1 GiB; longer contexts can require more. Command verified from upstream documentation, not freshly executed here.
Model information sources
$ how it compares
The decision models ranked either side of Jebadiah 27B v2.1 (Ollaya Q4_K_M), out of 357 with a score, then more from frontier-infra.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 123 |  Xor 1.2 | 59±1 | model files | ||
| 124 | jev-d3-46k | 59±4 | model files | ||
| 125 | Jebadiah 27B v2.1 (Ollaya Q4_K_M) | 59±4 | local | ||
| 126 |  vLLM-SR d3-mini (4B) | 58±1 | local | ||
| 127 |  Seb-9B v2 | 58±1 | model files | ||
| more from frontier-infra | |||||
| 57 | Jebadiah 27B v2.1 (T1e) | 65±1 | local | ||
| 143 | Jebadiah 9B v2.1 (T1e) | 57±1 | local | ||
| 160 | Jebadiah 9B v2.1 (Ollaya Q8_0) | 55±3 | local | ||
see all decision models · find the ones that fit your machine
$ what it's good at
knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▂ #94 of 263 · 45% of its score · 5 tests
Beats StartLux-Decision v1.1 4B, KnowLine 4B Gen5.5, Jeffan-4B-2010 here, though they rank higher overall. Trails Solar Decide, Clef-flash, JPT-35B-A3B here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Ollaya · PubMedQA | 80.8% ✎ | ▃▂▃▃·▂▃▃▆█▆▃ beats 25/25 | 16% |
| Ollaya · BoolQ | 89.7% ✎ | ▁··▃▃▁▄▁▃█▇▄ beats 25/25 | 12% |
| Ollaya · SQuAD 2 answerability | 86.6% ✎ | ▁▁▃▁▇█▃▃▄▃·▁ beats 22/25 | 6.5% |
| Ollaya · VitaminC verification | 80.3% ✎ | ▁···▃·▄█▃··▁ beats 21/25 | 6.3% |
| Ollaya · MultiNLI | 87.0% ✎ | ▂··▄██▂▆▆█▆▂ beats 12/25 | 3.9% |
workflows▁▁▁▂▁▂▂▆█▄▁▁ #44 of 114 · 50% of its score · 4 tests
Beats Clef-omni here, though they rank higher overall. Trails StartLux-Decision v1.1 2B, Solar Decide, openjev (DecisionBench ec4ff3df) here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Ollaya · SummEval relevance | 52.1% ✎ | ▁··▁▂▁▅··▃██ beats 24/25 | 17% |
| Ollaya · PAWS paraphrase | 82.8% ✎ | ▂▂▂▂·▃█▆·▂·▁ beats 17/25 | 14% |
| Ollaya · Aegis 2 safety | 81.2% ✎ | ▂·▁▁▂·▃▃▂█▃▁ beats 20/25 | 12% |
| Ollaya · SummEval consistency | 83.3% ✎ | ▁·▁▁··▁▁▁▁▄█ beats 19/25 | 6.3% |
$ where to run it
Other ways to run it
- Run it as published · needs 16.3 GB, NVIDIA only · from github.com/ollaya-dev
ollaya run jeb:27b --preset triage "The customer requests a refund."
Download model files ↗
Public ungated original GGUF verified at the exact benchmark manifest revision. Install the Ollaya runtime separately using its official installation instructions. Source · checked 2026-10-31.
Model repository and documentation: huggingface.co/frontier-infra ↗
$ how the score was worked out · Ollaya · original runs 95%
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 Ollaya · original runs | 9 | 95% |
| prior · pull toward the middle | 5% | |
| weighted average | = 58.5 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json