Skip to content
All releases
Added

Bison 1

Base Labs' top-ranked decision model joins the catalog with 128K context, text and image input, and a BYOK System One route through Baseten.

baselabs/bison-1 is a decision model from Base Labs, post-trained from Gemma 4 31B. It does not write chat text. Give it a state and a set of choice, yes/no or score questions and it returns calibrated option probabilities. It ranks first of 19 decision models on the Artificial Analysis Decision Index v2.1 at 76 and has the best calibration on the index (Brier 0.14). It accepts text and images over a 128K context.

It is BYOK-only on Baseten's native System One API. Add a Baseten key under Dashboard → BYOK; Baseten bills your own account and AnyRouter charges nothing for the route. Baseten lists it at $0.10 per 1M input tokens.

Decisions CLI
anyr decision --model "baselabs/bison-1"

Playbooks: Decisions API, AI SDK, playground.

Bar chart of Artificial Analysis Decision Index v2.1 scores: bison-1 76, d1.2-ultra 75, mammoth-1 71
Bison 1 ranks first of 19 decision models on the independent decision index v2.1. Vendor-reported.. Source: Baseten / Base Labs.
Scatter plot of decision-index accuracy against time per task, with bison-1, mammoth-1 and elephant-1 on the Pareto frontier
Accuracy against time per task for the Base Labs decision family. Vendor-reported.. Source: Baseten / Base Labs.

Looking for the full catalog? Browse all models.