{"slug": "base-labs-releases-top-ranked-decision-models-mammoth-1-and-and-elephant-1", "title": "Base Labs releases top-ranked decision models mammoth-1 and and elephant-1", "summary": "Base Labs released two open-source decision models, mammoth-1 (27B) and elephant-1 (9B), post-trained from Qwen3.5-27B and Qwen3.5-9B on 6.21 million decision-making examples, with mammoth-1 ranking #1 out of 208 decision models on the Jev Decision Index 0.4 and elephant-1 tied for #8 overall and the leading 9B decision model. On the separate Artificial Analysis Decision Index, elephant-1 is tied for #1 with the previous best model, d1-ultra. Both models return probability distributions over choice, score and boolean answers rather than generating text, and are free to try for a week on Baseten Model APIs with zero data retention by default.", "body_md": "### Published\n\nWe’re proud to release two of the world’s best decision models, trained by our own research lab: [mammoth-1](https://www.baseten.co/api/draft-mode/?url=/library/mammoth-1/&token=qea4cap5quz.MQJ1ebc) (27B) and [elephant-1](https://www.baseten.co/api/draft-mode/?url=/library/elephant-1/&token=qea4cap5quz.MQJ1ebc) (9B)! \n\nmammoth-1 ranks #1 out of 208 decision models, and elephant-1 is the leading 9B decision model on the Jev Decision Index. Both models are [open-source](https://huggingface.co/baselabs) and free to try for a week on Baseten Model APIs, with zero data retention (ZDR) by default.\n\nSince TypeSafe introduced Jev and the System One API, decision models have emerged as a new category. We post-trained Qwen3.5-9B and Qwen3.5-27B to build two of the best. Here’s a closer look at their capabilities, use cases, and how to evaluate them for your stack.\n\n## **What are decision models?** \n\nUnlike generative LLMs, decision models don’t generate text. Instead, they return a probability distribution (a score for every possible answer, with all scores summing to 1) over the possible answers to each question.\n\nYou can ask decision models 3 types of questions:\n\n- **Choice** : pick an option from a list (e.g., which team owns this ticket: infrastructure, billing, security, or support?)\n- **Score:** rate the input on a given scale (a 2 point scale would be low vs high). For example, how urgent is this ticket, low or high?\n- **Boolean** : answer yes or no (e.g., is this email a phishing attempt?)\n\n## **Where do decision models beat generative LLMs?**\n\nUsing generative LLMs for classification tasks, like routing a ticket or flagging spam, is costly and slow because the amount of computation they perform is overkill for the required output of the task. The response alone also doesn’t provide a reliable measure of certainty.\n\nDecision models like elephant-1 and mammoth-1 skip text generation, which makes it cheaper and faster to serve in production. Customizable confidence thresholds help determine when to accept an answer and when to escalate it. For example, a support ticket can be routed automatically when the model is 90% confident, and sent to a person when confidence is lower. The models can also be post-trained so their confidence better reflects how often they’re right.\n\nIn short, decision models are built to understand the request and make a decision without spending time writing a response. Compared to a reasoning model, a decision model may give up to a 10x speed up.\n\nHere are a few use cases:\n\n- Ticket triage and routing\n- Spam and phishing detection\n- Prompt guardrails and jailbreak detection\n- Routing requests between small and frontier LLMs\n\n## What elephant-1 and mammoth-1 are built for\n\nWe built elephant-1 and mammoth-1 to be generalist decision models that could handle a broad range of tasks, from routing support tickets to checking requests for safety. We trained and evaluated decision-making using data from five domains, mirroring previous work: knowledge and reasoning, language, tools and automation, retrieval and classification, and arts and human taste.\n\nmammoth-1 ranks #1 on the Jev Decision Index 0.4, while elephant-1 is tied for #8 overall and is the leading 9B decision model. On the separate Artificial Analysis Decision Index, elephant-1 is tied for #1 with the previous best model, d1-ultra.\n\nmammoth is a better fit if your task needs more general knowledge. If you mainly need quick reasoning, then choose elephant-1.\n\n## **Architecture and training** \n\nWe fine-tuned Qwen3.5-9B and Qwen3.5-27B on 6.21 million decision-making examples. To train each model, we used a dataset that combined public datasets with synthetically generated examples covering core decision skills, safety and abstention, and judging and expert domains. We refined the data through repeated training and evaluation runs with elephant-1, then used the final mix to train both models. This included examples that taught it when to answer “I don’t know,” when the model was uncertain about the accuracy of its answer. Read more about how we trained it in our technical report [here](https://labs.baseten.co/articles/mammoth-1-and-elephant-1-frontier-decision-models).\n\n## **How to choose the right decision model**\n\nelephant-1 and mammoth-1 are part of a growing set of decision models. To choose the right decision model for your workflow, we recommend you evaluate:\n\n- Accuracy: Does it make the right decisions? Include difficult cases.\n- Confidence: Does it know when it’s unsure? Check when to accept an answer and when to escalate, and set your confidence threshold accordingly.\n- Speed and cost: Measure latency and cost per decision at your expected traffic level. Include retries and fallbacks.\n\n## **Try elephant-1 and mammoth-1 today**\n\nYou can try [elephant-1](https://www.baseten.co/api/draft-mode/?url=/library/elephant-1/&token=qea4cap5quz.MQJ1ebc) and [mammoth-1](https://www.baseten.co/api/draft-mode/?url=/library/elephant-1/&token=qea4cap5quz.MQJ1ebc) for free for a week through Baseten Model APIs. Both models are also open-source on [Hugging Face](https://huggingface.co/baselabs). You can read the technical report from our research lab, Base Labs, [here](https://labs.baseten.co/articles/mammoth-1-and-elephant-1-frontier-decision-models). We look forward to seeing how you use them!", "url": "https://wpnews.pro/news/base-labs-releases-top-ranked-decision-models-mammoth-1-and-and-elephant-1", "canonical_source": "https://www.baseten.co/blog/the-best-decision-models-in-the-world-mammoth-1-and-elephant-1/", "published_at": "2026-10-21 17:27:45+00:00", "updated_at": "2026-10-21 17:42:27.384226+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "large-language-models", "ai-products", "ai-tools"], "entities": ["Base Labs", "mammoth-1", "elephant-1", "Qwen3.5-27B", "Qwen3.5-9B", "Jev Decision Index", "Artificial Analysis Decision Index", "Baseten Model APIs"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/base-labs-releases-top-ranked-decision-models-mammoth-1-and-and-elephant-1", "markdown": "https://wpnews.pro/news/base-labs-releases-top-ranked-decision-models-mammoth-1-and-and-elephant-1.md", "text": "https://wpnews.pro/news/base-labs-releases-top-ranked-decision-models-mammoth-1-and-and-elephant-1.txt", "jsonld": "https://wpnews.pro/news/base-labs-releases-top-ranked-decision-models-mammoth-1-and-and-elephant-1.jsonld"}}