# v1.0.19 Release Notes

> v1.0.18 -> v1.0.19

## Summary

Separates model choice from reasoning configuration and improves model-aware routing, while refining session visibility and chat controls. A new official guide and updated model tiers align the system with newer model capabilities, and related requests now keep the previous model to reuse the provider cache.

<details>
<summary>翻譯</summary>

明確拆分模型選擇與 reasoning 設定，並改善模型路由、工作階段識別與聊天介面。新增官方指南並更新模型分級，讓系統對齊新模型能力；同話題的請求會沿用上一輪模型以重用供應商快取。

</details>

## Changes

### FEAT

- Separate model selection from reasoning configuration (@pardnchiu) [ef7a8d9]
- Display session IDs and refine the TUI model status layout (@pardnchiu) [a9e32ef]

<details>
<summary>翻譯</summary>

- 明確拆分模型選擇與 reasoning 設定
- 顯示工作階段 ID 並調整 TUI 模型狀態版面

</details>

### UPDATE

- Reclassify GPT model tiers and add the GPT-6 guide (@pardnchiu) [b57c940]
- Rank model candidates by tier and reuse the previous model for related requests to improve cache reuse (@pardnchiu) [75cb9f5]

<details>
<summary>翻譯</summary>

- 重新調整 GPT 模型分級並新增 GPT-6 指南
- 依分級排序候選模型，並在同話題的請求沿用上一輪模型以提升快取命中

</details>

## Scope

- `internal/agents/exec/`, `internal/session/config/` — FEAT, UPDATE
- `configs/prompts/` — UPDATE
- `internal/runtime/tui/` — FEAT
- `page/` — UPDATE

---

Generated by [SKILL](https://github.com/pardnchiu/skill-version-generate)