← Back to models
Base Labs logo

Specs

Input price
On demand
Output price
On demand
Context window
16.4K tokens

Base Labs' 27B open-weight decision model - an MIT-licensed Qwen3.5-27B fine-tune that returns calibrated probabilities over a fixed question set in one forward pass.

Capabilities

  • 27B parameters, full fine-tune of Qwen3.5-27B, MIT licensed
  • Typed questions (`noul`, `choice`, `score`) answered from one shared-prefix prefill
  • Brier calibration term and a trained "I don't know" option
  • Top score on the Jev Decision Index 0.4 at 71.24 in Base Labs' release
  • 16,384-token sequence limit with hard refusal, not silent truncation

Best for

  • Knowledge-heavy decisions that depend on facts beyond the input text
  • Routing and classification where a calibrated probability drives the threshold
  • Workflows that need a confidence score before escalating to a human reviewer

Limitations to keep in mind

  • 16K context is a short document; longer inputs are refused
  • Text only - scanned faxes need OCR first
  • Decision benchmarks are not clinical validation
  • The MIT fine-tune sits on Qwen3.5 base weights with their own terms
  • Self-hosting needs a CUDA GPU and a pinned OneJev runtime

HIPAA-compliant hosting

mammoth-1 is available under our HIPAA-compliant design-partner program, with a signed Business Associate Agreement (BAA), encryption in transit and at rest, and access controls. We are currently onboarding design partners, with general availability coming soon. Run the model behind a unified OpenRouter-style API and swap to another model with a single parameter. For the deployment requirements behind that, see our 2026 guide to HIPAA compliance for AI & LLM APIs.

Pricing in context

Open source models like mammoth-1 can be served efficiently on optimized inference infrastructure, with savings passed through to you. Exact savings depend on the model and your volume, but open source inference is typically a fraction of the per-token cost of closed-source frontier models on Azure OpenAI or AWS Bedrock - without cloud egress lock-in or minimum commitments. See the full pricing comparison or browse the model catalog.

Related models

Design partner program

Run mammoth-1 under HIPAA

Join the waitlist to be prioritized. We'll reach out with a qualification call and early access.

Join the waitlist