Model vibe run, July 2026

71 prompt results from one run on one model, Gemini 3 Flash Preview, on 16 July 2026. Responses are shown as generated; no scores were assigned.

This run covers one model only: Gemini 3 Flash Preview. The run was set up for five Google models. Four had been retired or were unavailable to new users when it started:

• gemini-2.5-flash: This model models/gemini-2.5-flash is no longer available to new users. Please update your code to use a newer model for the latest features and improvements.
• gemini-2.0-flash: This model models/gemini-2.0-flash is no longer available. Please update your code to use a newer model for the latest features and improvements.
• gemini-2.0-flash-lite: This model models/gemini-2.0-flash-lite is no longer available. Please update your code to use a newer model for the latest features and improvements.
• gemini-2.5-flash-lite: This model models/gemini-2.5-flash-lite is no longer available to new users. Please update your code to use a newer model for the latest features and improvements.

All prompts used the workbench's text-generation route. Image-generation prompts did not invoke image generation, and agentic/research prompts did not receive tools. Their text responses are archived as generated, without treating them as successful image or agent executions.

Six attachment-backed prompts ran without their attachments. The workbench rejected three document paths as outside its allowed attachment directory; the three image prompts were executed with a model configured as not supporting vision, so those images were skipped. The provider returned responses, but these six outputs are not valid attachment-grounded evaluations.

Run totals: 67 prompts, 71 results (some prompts have more than one step), 0 errors.