The L&S AI Inference platform provides three models, each optimized for different tasks. Picking the right one matters — using the wrong model for a task can mean slower responses, lower quality output, or unnecessary load on shared resources.
Best for: Everyday questions, writing assistance, summarization, brainstorming, drafting, Q&A
Strengths:
Limitations:
Use it when: You want a quick, clear answer or need help with text-based tasks. If you're not sure which model to use, start here.
Best for: Code generation, debugging, refactoring, scripting, tool use, agentic coding workflows
Strengths:
Limitations:
Use it when: You're writing, debugging, or refactoring code; working with scripts or config files; or running agentic coding tasks. For straightforward coding work, this model will be faster and equally good as gpt-oss-120b.
Best for: Complex analysis, long documents, multi-step problems, research synthesis, nuanced comparisons
Strengths:
Limitations:
qwen3-coder-next on straightforward coding tasksUse it when: The task genuinely requires deep reasoning, long-context understanding, or careful synthesis. Don't default to it just because it's the biggest — for many tasks, gemma-4-31b or qwen3-coder-next will be faster and equally good.
| Task | Recommended model | Why |
|---|---|---|
| 2-page abstract | gemma-4-31b | Fast, conversational, handles short summaries well |
| 30-page paper | gpt-oss-120b | Needs to hold more context; benefits from deeper synthesis |
| Task | Recommended model | Why |
|---|---|---|
| Draft an email | gemma-4-31b | Fast, natural writing |
| Write a grant proposal | gpt-oss-120b | Benefits from structured multi-section reasoning |
| Literature review outline | gpt-oss-120b | Complex synthesis task |
| Proofread a paragraph | gemma-4-31b | Simple editing task, no need for the larger model |
| Compare three competing theories | gpt-oss-120b | Nuanced multi-perspective analysis |
| Task | Recommended model | Why |
|---|---|---|
| Write a Python script | qwen3-coder-next | Purpose-built for code generation |
| Debug a function | qwen3-coder-next | Strong at identifying and fixing errors |
| Refactor a large file | qwen3-coder-next | 256k context handles large files well |
| Explain what a script does | gemma-4-31b | Simple explanation task; no need for a coding model |
| Architect a multi-service system | gpt-oss-120b | Benefits from deep reasoning across many constraints |
Questions or issues? Contact [help@cit.ucsb.edu].