Tasks and inputs
Drex 1.5 is listed for Structured Decisions. It accepts text and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
Drex 1.5 evaluates typed questions against a shared state and returns calibrated probabilities through the System One API.
Pricing conditions: Drex 1.5 costs $0.05 per million input tokens; output tokens are not billed. Its 131,072-token state limit and 139,264-token row limit are distinct. It remains the hosted fallback for requests too long for the Drex 2.0 fleet.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
Drex 1.5 is a specialized model for evaluating typed questions against a shared state. It returns probabilities that an application can use in a decision workflow. This is a different interface from asking a chat model to write a paragraph: question definitions, calibration and the threshold for acting on an answer matter as much as the input text.
The publisher lists drex-v1.5 as the version released on 28 September 2026. Requests go to the System One endpoint. The drex-latest alias moved to Drex 2.0 on 12 October 2026, while the explicit drex-v1.5 ID remains available; a pinned version and a moving alias have different reproducibility properties. The response identifies the model that answered the request. Drex 1.1 has retired and is served by a newer version under the documented replacement rules.
Drex lists a 131,072-token state limit and a 139,264-token row limit. The row includes the state plus the longest question. Neither number should be presented as a chat model’s maximum generated answer length. The model page lists $0.05 per million input tokens; confirm the current billing rules for the request shape you use.
Keep a held-out set of cases with known outcomes. Inspect calibration, false positives and false negatives at the thresholds your application proposes to use. Include ambiguous questions and examples outside the expected distribution. Changing a question’s wording or the structure of the state can change the decision, so version those alongside the model.
The source links document the interface and limits. SyncDev does not claim an independent calibration measurement.
Compare the model on the work you need to complete.
Drex 1.5 is listed for Structured Decisions. It accepts text and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
Documented limits: 131.07K context. The listed route has starting rates of $0.05 input and $0.00 output per million tokens. Measure a completed task using the prompts, documents and tools your application will use. Include reasoning and retries in the result.
Use the model identifier and account requirements for the selected API route. Confirm its request limits and output format with one complete task before connecting a larger workflow.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
phocinae
Phocinae Largha 150M v1 from phocinae: Encoder decision model that answers typed questions (noul, choice, score) over text with calibrated probabilities, on CPU and GPU.
Nace AI
Drex 2.0 scores typed questions over text and returns probabilities. Its hosted route uses 2.0 for short states and falls back to 1.5 for longer requests.
Liquid AI
Liquid d1 returns typed binary decisions, class distributions or scores through Liquid AI’s decision API.
TypeSafe
TypeSafe’s Jev evaluates many questions against shared text state through a System One interface; the latest alias currently resolves to Jev 1.13.