LLM Gemini Plugin 0.33 Adds Gemini 3.7 Flash and Server-Side Tool Execution
Simon Willison released version 0.33 of the llm-gemini plugin, introducing support for Google's Gemini 3.7 Flash and Gemini 3.5 Flash-Lite. The release enables CLI developers to inspect reasoning traces and run server-side code execution tools directly from the terminal.
Impact: Medium
Why it matters
Run Gemini 3.7 Flash directly in terminal scripting pipelines with native reasoning traces and sandboxed Python code execution.
TL;DR
- 01Install or update llm-gemini to 0.33 to access Gemini 3.7 Flash in the LLM CLI.
- 02Enable server-side sandboxed execution with `-T CodeExecution` for deterministic calculations.
- 03Inspect full model reasoning traces directly within terminal workflows.
Key facts
- Plugin Version
- 0.33
- Supported Models
- gemini-3.7-flash, gemini-3.6-flash, gemini-3.5-flash-lite
- Embedding Models
- gemini-embedding-2, gemini-embedding-001
Gemini 3.7 Integration for the LLM CLI
Simon Willison updated the llm-gemini plugin to version 0.33, bringing Google's latest model family directly into terminal workflows. The release supports gemini-3.7-flash, gemini-3.6-flash, and gemini-3.5-flash-lite, along with gemini-embedding-2 and gemini-embedding-001.
Server-Side Code Execution and Reasoning
Built for compatibility with LLM 0.32, the plugin surfaces internal reasoning traces from Gemini models. Developers can also enable server-side tools directly from the command line using the -T flag.
For example, passing -T CodeExecution allows Gemini 3.7 Flash to execute Python scripts in Google's cloud sandbox to compute precise mathematical results before returning the final response.
Try it in 2 minutes
llm -m gemini-3.7-flash -T CodeExecution \
'use python to calculate (factorial of 13) * 3'bash
✓ When to use
- Automating terminal-based workflows with fast reasoning models
- Executing deterministic math or Python scripts via Google-hosted tools
✕ When NOT to use
- Offline local inference without internet access or API keys
- Environments requiring air-gapped data residency without third-party tool execution
What to do today
- Run `llm install -U llm-gemini` to update the plugin.
- Test tool execution with `llm -m gemini-3.7-flash -T CodeExecution '...'`.
Sources