Quyet Models by Chinh Nguyen

Chinh NguyenApache 2.0Open sourceAgent
Chinh Nguyen releases · 2 in the last 12 monthsChangelog →
2 models2026Up to 8k ctx

Last refreshed 2026-10-05. Next refresh: weekly.

Details

ResearcherChinh Nguyen
LicenseApache 2.0OSI-approved
Commercial useCommercial use: permitted
Models2
Released2026
Max context8k

Capabilities

Structured OutputsAll models

Links

WebsiteHuggingFace

About

Quyet is Chinh Nguyen's open-weight family of calibrated decision models. Each takes a state (text, JSON, or a conversation) and typed questions (choice, score, or true/false 'noul'), picks one option per question, and returns calibrated probabilities, using the TypeSafe /v1/systemone answer shape. Quyet 1.0 (October 2026) ships in Large (Gemma-4-31B-it decoder), Medium (Qwen3.5-4B decoder), and Small, Small-EN, and Tiny (encoder) sizes under Apache 2.0, served through the quyet Python package. The family is tuned for English and Vietnamese.

Decision facts

Best fit
agentstructured outputscoding
Capability starting point
Quyet-1.0-Large with 8k context and structured outputs
Lowest tracked input
Not tracked
Closest related family
Torchcast Decision

Current Variants

Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.

2 in view
Quyet-1.0-LargeCurrent

Use when the workload needs 8k context, 31.3B parameters, and structured outputs.

2026-108k context31.3B parametersstructured outputs
Quyet-1.0-MediumCurrent

Use when the workload needs 8k context, 4.7B parameters, and structured outputs.

2026-108k context4.7B parametersstructured outputs

Release Timeline

1 release group
2026-10
2 current
Quyet-1.0-Large
8k context31.3B parametersstructured outputs
Current
Quyet-1.0-Medium
8k context4.7B parametersstructured outputs
Current

Specifications(2 models)

Quyet model specifications comparison
ModelReleasedContextParametersStructured Outputs
Quyet-1.0-Large2026-108k31.3BYes
Quyet-1.0-Medium2026-108k4.66BYes