You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
description: Adds Moonshot kimi-k2-thinking and MiniMax-M2-Stable with prompt caching, smoother mode import flow, and multiple UI and reliability fixes.
3
+
keywords:
4
+
- roo code 3.30.3
5
+
- new features
6
+
- bug fixes
7
+
- provider updates
8
+
image: /img/v3.30.3/v3.30.3.png
9
+
---
10
+
11
+
# Roo Code 3.30.3 Release Notes (2025-11-07)
12
+
13
+
This release adds support for Moonshot's newest model, kimi‑k2‑thinking, and includes several UI improvements and a few bug fixes.
## New models and caching: Moonshot kimi‑k2‑thinking and MiniMax‑M2‑Stable
18
+
19
+
kimi‑k2‑thinking: Moonshot's latest & best‑performing model. Plus MiniMax‑M2‑Stable with prompt caching to cut latency and cost on repeated prompts (thanks nokaka!) ([#9079](https://github.com/RooCodeInc/Roo-Code/pull/9079), [#9072](https://github.com/RooCodeInc/Roo-Code/pull/9072)).
20
+
21
+
## UI and workflow improvements
22
+
23
+
We’ve batched several UX updates to make everyday use smoother ([#9057](https://github.com/RooCodeInc/Roo-Code/pull/9057), [#8932](https://github.com/RooCodeInc/Roo-Code/pull/8932), [#8965](https://github.com/RooCodeInc/Roo-Code/pull/8965), [#9003](https://github.com/RooCodeInc/Roo-Code/pull/9003)):
24
+
25
+
- Home screen: cleaner left‑aligned layout, clearer recent history (up to 4), tips auto‑dismiss after 6 tasks, improved spacing on large windows.
26
+
- Diff view in chat: unified format with line numbers and inline added/removed counts for faster, more accurate reviews.
27
+
- Settings clarity: setting Error & Repetition Limit to 0 explicitly disables that safety mechanism, avoiding unexpected auto‑stops.
28
+
- Mode import flow: after importing a mode, Roo auto‑switches to it; if the mode data hasn’t loaded yet, Roo briefly falls back to Architect mode to avoid UI desync.
29
+
30
+
Together these reduce clicks, improve readability, and make behavior more predictable.
31
+
32
+
## Bug Fixes
33
+
34
+
* Auto‑retry on empty assistant response to prevent task cancellations (reporter: Akillatech) ([#9083](https://github.com/RooCodeInc/Roo-Code/pull/9083))
35
+
* Use “system” role for non‑streaming OpenAI‑compatible requests to prevent provider errors (reporter: whitfin) ([#8216](https://github.com/RooCodeInc/Roo-Code/pull/8216))
36
+
* Prevent notification sound when messages are queued to auto‑continue ([#8540](https://github.com/RooCodeInc/Roo-Code/pull/8540))
Copy file name to clipboardExpand all lines: docs/update-notes/v3.30.mdx
+9-1Lines changed: 9 additions & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -25,6 +25,9 @@ OpenRouter currently supports 7 embedding models, including the top‑ranking Qw
25
25
26
26
* Terminal settings cleanup with Inline as the default terminal and clearer options; shell integration default is disabled to reduce environment conflicts ([#8342](https://github.com/RooCodeInc/Roo-Code/pull/8342))
27
27
* File path tooltips show full paths on hover across reads/edits, directory listings, and search, reducing confusion in narrow layouts (thanks da2ce7!) ([#9030](https://github.com/RooCodeInc/Roo-Code/pull/9030))
28
+
* Home screen visuals refined for clarity and space usage ([#9057](https://github.com/RooCodeInc/Roo-Code/pull/9057))
29
+
* Chat diff appearance is clearer and easier to scan ([#8932](https://github.com/RooCodeInc/Roo-Code/pull/8932))
30
+
* Settings copy clarifies that an Error & Repetition Limit of 0 disables the mechanism ([#8965](https://github.com/RooCodeInc/Roo-Code/pull/8965))
28
31
29
32
## Bug Fixes
30
33
@@ -41,6 +44,9 @@ OpenRouter currently supports 7 embedding models, including the top‑ranking Qw
41
44
* Restore correct 32K maxTokens for Claude Opus 4.1 to avoid premature truncation (thanks kaveh-deriv!) ([#9046](https://github.com/RooCodeInc/Roo-Code/pull/9046))
42
45
* Fix dynamic provider model validation so provider switches use a valid default model without TypeErrors (thanks NotADev137!) ([#9054](https://github.com/RooCodeInc/Roo-Code/pull/9054))
43
46
* Include full user agent details in AWS Bedrock requests for better auditing and troubleshooting (thanks ajjuaire!) ([#9043](https://github.com/RooCodeInc/Roo-Code/pull/9043))
47
+
* Auto‑retry on empty assistant response to prevent task cancellations (reporter: Akillatech) ([#9083](https://github.com/RooCodeInc/Roo-Code/pull/9083))
48
+
* Use “system” role for non‑streaming OpenAI‑compatible requests to prevent provider errors (reporter: whitfin) ([#8216](https://github.com/RooCodeInc/Roo-Code/pull/8216))
49
+
* Prevent notification sound when messages are queued to auto‑continue ([#8540](https://github.com/RooCodeInc/Roo-Code/pull/8540))
44
50
45
51
## Misc Improvements
46
52
@@ -56,4 +62,6 @@ OpenRouter currently supports 7 embedding models, including the top‑ranking Qw
56
62
* Fireworks: add MiniMax M2 with 204.8K context and 4K output tokens; correct pricing metadata (thanks dmarkey!) ([#8962](https://github.com/RooCodeInc/Roo-Code/pull/8962))
57
63
* Set Claude Sonnet 4.5 as the default where available across Anthropic, Bedrock, OpenRouter, Claude Code, and Unbound ([#8922](https://github.com/RooCodeInc/Roo-Code/pull/8922))
58
64
* Cerebras zai‑glm‑4.6: 40,960 max output tokens and a 131,072‑token context window; faster responses (~1,000 tok/s) (thanks sebastiand-cerebras!) ([#9024](https://github.com/RooCodeInc/Roo-Code/pull/9024))
59
-
* OpenRouter: add Qwen3‑Embedding‑0.6B (1024‑dim) and 4B (2560‑dim) for code indexing (thanks dmarkey!) ([#9060](https://github.com/RooCodeInc/Roo-Code/pull/9060))
65
+
* OpenRouter: add Qwen3‑Embedding‑0.6B (1024‑dim) and 4B (2560‑dim) for code indexing (thanks dmarkey!) ([#9060](https://github.com/RooCodeInc/Roo-Code/pull/9060))
0 commit comments