Vicuna-13B

About

The Vicuna Team introduces Vicuna-13B, an open-source chatbot. It is trained by fine-tuning LLaMA on user-shared conversations collected from ShareGPT. The vendor states that preliminary evaluation using GPT-4 as a judge shows Vicuna-13B achieves more than 90% quality of OpenAI ChatGPT and Google Bard, and outperforms LLaMA and Stanford Alpaca in more than 90% of cases. The training cost for Vicuna-13B is around $300. The code and weights, along with an online demo, are publicly available for non-commercial use. The page describes that Vicuna becomes capable of generating more detailed and well-structured answers compared to Alpaca after fine-tuning with 70K user-shared ChatGPT conversations. The training process involved enhancing Alpaca's training scripts to handle multi-turn conversations and long sequences, expanding the max context length to 2048, and utilizing gradient checkpointing and flash attention for memory optimization. The vendor also used SkyPilot managed spot instances to reduce training costs.

Read from lmsys.org on 2026-08-26. We describe what the vendor states; we do not audit it.

Composite of public compliance, deployment, integration, adoption, and company signals. “Not disclosed” reflects gaps in available data, not a vendor deficiency.

Enterprise Use Cases

Integrations

How Vicuna-13B compares

Peers from the same category, ranked by Enterprise Readiness. Readiness is a composite of cataloged compliance, deployment, integration, adoption, and company signals.

Frequently Asked Questions