--- format: "aidr-story-markdown/v1" id: "7a746d592683e9c32a15276c21f89204a69a2c4927ffb61265a69f59ad17733e" canonical_url: "https://aidr.today/7a746d59?lang=en" title: "Bespoke Nimble Hits 90% Accuracy in First Open Source Jev Model" lang: "en" requested_lang: "en" available_langs: ["en","vi"] translation_fallback: null fallback_fields: [] published_at: "2026-09-19T08:47:33.000Z" category: "Models" topics: ["bespoke","nimble","qwen","fine-tuning","open-source","decision-model","lora","bespoke-labs"] source_urls: ["https://marketbrief.now/ai/bespoke-nimble-hits-90percent-accuracy-in-first-open-source-jev-model-427e9677","https://huggingnews.com/ai/bespoke-nimble-hits-90percent-accuracy-in-first-open-source-jev-model-427e9677","https://x.com/madiator/status/2100990591215783946","https://x.com/bespokelabsai/status/2100995803540255030","https://x.com/suchenzang/status/2101196249726611620","https://x.com/Tim_Dettmers/status/2101212492009476143","https://huggingnews.com/ai/bespoke-labs-releases-nimble-open-model-that-hits-90percent-on-jev-eval-9d23dce9","https://huggingnews.com/ai/update-bespoke-labs-builds-open-jev-rival-in-2-days-using-9b-model-cc9dd3f5"] summary: "A LoRA finetune of Qwen3.5-9B served as the foundation for Nimble, a new decision model released by Bespoke with its full data and training recipe. The model u…" --- # Bespoke Nimble Hits 90% Accuracy in First Open Source Jev Model > [Open the canonical story]() **Published:** 2026-09-19T08:47:33.000Z **Category:** Models **Topics:** bespoke, nimble, qwen, fine\-tuning, open\-source, decision\-model, lora, bespoke\-labs ## Summary A LoRA finetune of Qwen3\.5\-9B served as the foundation for Nimble, a new decision model released by Bespoke with its full data and training recipe\. The model u… ## Sources - [Story source]() - [Story source]() - [Story source]() - [Supporting source]() - [Supporting source]() - [Supporting source]() - [Story source]() - [Story source]()