# lib/llm — Logbook

Model additions, removals, and configuration changes. Pricing details: [conventions/llm.md](https://static.localhost/~/.myconf/conventions/llm.md) | [LLM_MODELS.md](https://static.localhost/~/.myconf/LLM_MODELS.md)

| Date       | Action  | Type  | Change                                                                       | Notes                                                               |
|------------|---------|-------|------------------------------------------------------------------------------|---------------------------------------------------------------------|
| 2026-02-26 | Upgrade | image | `gemini-flash`: `gemini-2.5-flash-image` → `gemini-3.1-flash-image-preview` | Nano Banana 2. Pro quality at Flash speed. Old kept as `gemini-flash-v1`. Now supports imageSize. |
| 2026-02-26 | Upgrade | text  | `gemini`: `gemini-3-pro-preview` → `gemini-3.1-pro-preview`                 | ARC-AGI-2 77.1% (vs 31.1%). Same pricing. Consumers switched to aliases. Both API + subscription work. |
| 2026-02-26 | Add     | image | `gpt-image-mini`: `gpt-image-1-mini`                                        | $0.011/img (med, 1K). 74% cheaper than gpt-image-1.5. Fixed response_format bug for GPT Image models. |
| 2026-02-26 | Note    | text  | Deprecation audit: upcoming shutdowns                                        | `gemini-3-pro-preview` Mar 9. `gemini-2.5-pro/flash` Jun 17. `gemini-2.0-flash` Jun 1. `flash-lite` Jul 22. No 3.1 Flash text model yet. |