We are updating the UCSB AI Commons platform to provide increased capacity, more clarity around resource usage, and bring you access to the latest generation of AI models.
Transitioning to a "Credit Budget"
To give you a clearer understanding of your resource availability, we are shifting our terminology from "token budget" to "credit budget." As part of this shift, we will be increasing your budget to 35 million monthly credits per user.
- How it works: Different AI models have varying operational costs. Instead of a flat token count, your allocation is now represented as credits. Higher-capability models draw more credits per query, while fast, lightweight models consume significantly fewer.
- Credit value: One credit is pegged to one token processed by our base workhorse model (Claude Haiku). The credit-to-token "exchange rate" appears in the model selection drop-down menu.
- Tracking Usage: Your existing usage metrics remain intact. You can continue tracking your consumption in the dashboard using the same workflow as before.
New Models Available
Starting today, you have access to several new state-of-the-art models. As a reminder, if you need help choosing a model, you can ask the chatbot within the AI Commons.
- Claude 5.5 Opus, Sonnet, and Haiku. With this update, Haiku is an exceptionally fast, cost-effective option for everyday tasks and high-volume generation.
- Introducing the first frontier models from OpenAI on AI Commons: Sol 6 and Luna 6. These offer capabilities similar to Claude 5.5 Opus and Haiku.
- Nvidia Nemotron 3 Super 120B is a robust Open Source and open-weight model if you require an LLM that is less of a black box.
- Moonshot AI Kimi K3 is a mixture-of-experts open-weight model that offers capabilities close to Claude Opus 5.5.
Model Deprecations (Effective Nov. 6)
To keep the platform focused on top-tier performance and efficiency, the following models will be retired from the chat interface. If your workflow relies on these specific models, we recommend transitioning to one of the more recent LLM offerings.
- Amazon Nova Micro, Light, and Pro
- Meta Llama 4 Scout and Maverick
- Qwen 3
My AI Commons Portal Launched
We have launched My AI Commons, a new dedicated members area designed to streamline workflows and connect you directly with key platform resources.
- Current Features: Access platform announcements and changelogs, request extended token allocations, and book time with a Solution Architect.
- Coming Soon: Technical blog, community contributions, and additional collaboration tools.
Model Capabilities & Technical Specs
For technical users interested in benchmark comparisons, architecture specs, and underlying model capabilities:
- View our Model Comparison Chart for side-by-side performance and credit-cost breakdowns.
- Access official Model Documentation to review detailed technical parameters for each provider.
Questions or feedback? Please open a support ticket.
News