mammoth-1 A 27B open-source decision model trained by Baseten's Base Labs research team. It ranks #1 on the Jev Decision Index. Released together with the 9B elephant-1 model, which is the top model under 15B parameters and tied for #1 on the Artificial Analysis Decision Index. Both have open weights and are available free for one week on Baseten Model APIs.
Visit productFeatured media
Loading media…
View on XCosigns Received
“Excellent model. We have included evaluations of two models in our medical reliability benchmark. website： https://medical-decision-atlas.pages.dev/”
“used elephant-1 to classify hacker news comments > 1500 comments > bucketed into sentiment around AI from enthusiast to doomer spot-checked a bunch. some learnings. > picking the categories and defining the criteria for each category is at least half the work when using a decision model > need to look at the distribution of the spread between the first- and second-ranked categories to make sure the model is truly confident > classic hn sarcasm makes this extra hard but also more fun”
“Our research lab @baselabs just trained the #1 decision model in the world. mammoth-1 (27B) is #1 of 208 on the Jev Decision Index. elephant-1 (9B) ties for #1 on @ArtificialAnlys' Decision Index. Open weights, free on @baseten until Oct 27. I tried it on an agent problem 🧵”