elythros-decider-0.6b
Qwen3-0.6B fine-tuned to return typed decisions, 596M parameters.
Community card
What this is
elythros-decider-0.6b is a typed-decision model: given a document and a closed set of options, it returns one of those options with a probability across the whole set rather than free text.
Its card lists training data: PolyAI/banking77, clinc/clinc_oos, nyu-mll/glue, stanfordnlp/snli, google/boolq, stanfordnlp/sst2, cais/mmlu. That matters when you read any benchmark number about it, because a model tuned on a benchmark's own training split has an advantage over one that has never seen it.
How to run it
MLX on Apple Silicon
MLX uses the unified memory on M-series chips, which is the fastest local option on a Mac. Full MLX on Apple Silicon guide.
mlx_lm.generate --model Elythros/elythros-decider-0.6bA public number is a shortlist, not a decision
The only benchmark that settles which model you ship is one built from your own historical decisions. We freeze that test set, run the candidates against it, and hand back the threshold your error budget supports.