Cline users who do not pin a model will get a different default after this release, and cloud tasks now have their own model setting. Retry behavior also drops from 5 attempts to 3.
Key takeaways · AI-distilled
Free-tier Gemini API keys failed every turn with a 429 quota error on Gemini 3+ because Cline's gemini provider sent native Google Search, and on Gemini 2.x every Cline tool was dropped. Cline 4.1.26 stops sending native Google Search on that provider.
GPT-6 models such as gpt-6-sol and gpt-6.1-sol now work on OpenAI and Azure OpenAI-compatible endpoints, and SAP AI Core stops rejecting requests over 'max_tokens', because Cline now sends max_completion_tokens to both.
Most providers get new default models in this release, mostly Kimi K3.1 Pro, MiniMax M3.1 Flash or DeepSeek V4.1 Flash 1017, and ClinePass now defaults to Kimi K3.1 Pro. The catalog adds CommandCode, Reflection, Liquid inferenceRunning a trained model to get answers — the phase where AI is actually used, as opposed to trained.Full definition → and Munito AI and drops CrofAI, iFlow and Llama.
Two long-task fixes: a turn whose connection drops after the model streamed only reasoning is now retried instead of ending the task, and auto-context compactionSummarizing an agent's earlier conversation to free room in the context window so a long session can keep going.Full definition → no longer runs for models with an unknown context windowThe maximum amount of text a model can consider at once — its working memory for the current conversation or task.Full definition → limit or re-runs every few tool calls after the first compaction.
Terms in this piece · Glossary
sandbox — An isolated environment where AI-generated code or agent actions run without being able to touch anything real.
inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
context compaction — Summarizing an agent's earlier conversation to free room in the context window so a long session can keep going.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
