mammoth-1 and elephant-1: Baseten's open-weight decision models top the Jev Decision Index
Baseten's internal research lab Base Labs released two open-weight decision models — mammoth-1 (27B) and elephant-1 (9B) — post-trained on Qwen3.5 to return calibrated probabilities instead of generated answers. mammoth-1 ranks #1 on the Jev Decision Index; both are on Hugging Face and free to try for a week.
Why it matters
A post-training team turned off-the-shelf Qwen3.5 into the world's top-ranked decision models with 6.21M training examples, showing open models can rival frontier labs on high-stakes structured decisions.
What you could build with it
An indie developer building a support-ticket triage SaaS could use elephant-1 to classify incoming tickets and return a calibrated confidence per routing decision, auto-routing only high-confidence cases and queuing borderline ones for human review — cutting manual triage time while keeping an auditable trail for every decision the system makes.
Does it hold up?
Too early to judge — released October 21, 2026, the benchmark lead is vendor-reported and no independent production deployments or re-tests have surfaced yet.
Built with mammoth-1 and elephant-1
- Baseten's Mammoth-1 and Elephant-1 now available on AI Gateway for freeVercel changelog · Vercel added baseten/mammoth-1 and baseten/elephant-1 to AI Gateway for free access through October 27, letting developers route decision calls through a unified API and track them alongside other model usage.
- Baseten's Base Labs turns Qwen into decision models with 6.21M examplesRuntimeWire analysis · Ryan Merket's analysis explains how Max Kirkby's Base Labs fine-tuned Qwen3.5 into specialist decision systems with 6.21M examples, and discusses why the benchmark claim is testable but real-workflow transfer matters more.
- Decision Index Full Scores — Jev leaderboardBenchLM leaderboard · The Jev Decision Index full score table (209 evaluated systems) shows mammoth-1 (27B) ranked #1 at 71.2 points, ahead of Perplexity Decider v1.2 and StartLux-Decision v1.1.
Learn more
First spotted on article: source.
More AI models
OpenAI publishes 722 AI-generated math manuscripts from an unreleased modelOpenAI released 722 mathematical manuscripts in 372 result families to GitHub under Apache-2.0, produced by an…models · JEV 0.74Tencent Hunyuan Hy3: open 295B MoE with 21B active under Apache 2.0Tencent open-sourced Hy3, a 295B-parameter mixture-of-experts model with only 21B parameters active per token, under…models · JEV 0.73NVIDIA Isaac Lab 3.0 GA + Isaac OS open runtime + Agile ONE SN humanoidMajor Isaac platform expansion at GTC Berlin Oct 21: Isaac Lab 3.0 is generally available as the open-source robot…models · JEV 0.73StepFun Step-5-Preview — open BF16 weights releasedStepFun published the full BF16 weights for its flagship Step-5-Preview agentic model (600B-parameter sparse MoE, 27B…models · JEV 0.73NVIDIA Nemotron 3.5 Retriever collection: Embed 8B/1B, Rerank 8B, Parse 631MNVIDIA's Oct 21 collection drop for agentic retrieval: Nemotron-3.5-Embed in 8B BF16, 1B BF16 and 1B NVFP4…models · JEV 0.72Kimi K3.1 Pro: Moonshot's 2.8T open-weight flagship officially launchesMoonshot AI officially launched Kimi K3.1 Pro on Oct 23, a mixture-of-experts model with 2.8 trillion total / 104…models · JEV 0.72
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.