Cline v3.88.0 Adds Fireworks AI Kimi K2.6 as Default Model, Fixes MCP Server Management
Cline, the AI coding assistant, released v3.88.0 on June 5, 2025, switching its default Fireworks AI model to Kimi K2.6. The update fixes critical MCP server management bugs and enables the upstream recommended models endpoint for all users.
Cline v3.88.0 Adds Fireworks AI Kimi K2.6 as Default Model, Fixes MCP Server Management
Cline released v3.88.0 on June 5, 2025, changing its default Fireworks AI model to Kimi K2.6 and addressing critical bugs in MCP server configuration.
Key Changes
The update adds the latest Fireworks AI serverless models to Cline's model selection and sets Kimi K2.6 as the new default Fireworks model. Cline removed outdated Fireworks AI models and corrected model metadata and cache pricing information.
The release fixes a significant bug in MCP (Model Context Protocol) server management where delete and add operations would cause the file watcher to empty the entire MCP server list during settings writes. This bug prevented users from reliably managing their MCP server configurations.
Recommended Models Endpoint
Cline now uses the upstream Cline recommended models endpoint for all users, removing the feature flag that previously gated this functionality. This change means all Cline instances will now receive standardized model recommendations directly from the project's maintained endpoint.
What This Means
The switch to Kimi K2.6 as the default Fireworks model suggests the Cline team found it performs better than their previous default for coding tasks. Moonshot AI's Kimi models are known for their extended context windows and Chinese-English bilingual capabilities.
The MCP server fix addresses a critical workflow issue for users who integrate external tools and data sources through the Model Context Protocol. The bug could have caused data loss in server configurations, making this a significant stability improvement.
Removing the feature flag for the recommended models endpoint indicates the Cline team is consolidating their model recommendation infrastructure, likely to provide more consistent experiences across their user base and simplify testing of new model integrations.
Full changelog available from cli-v3.0.20 to v3.88.0.
Related Articles
Google's Data Agent Kit reaches GA with 37 open-source skills and an IAM deny policy for read-only agents
Google has made Data Agent Kit generally available. It is free, open source, and installs into Antigravity, Claude Code, Codex, Cursor, and VS Code-compatible editors. It splits guardrails between advisory skills and enforceable IAM deny policies, according to InfoQ.
iFLYTEK launches Agent OS, APA and XFYUN CLI, putting 700-plus AI capabilities behind one command
At its 2026 Global 1024 Developer Festival in Beijing on October 22, iFLYTEK's open platform introduced Agent OS, APA, the XFYUN CLI and Xingchen TaaS (Token as a Solution). The XFYUN CLI lets one command call more than 700 atomic AI capabilities and more than 200 agent adaptation capabilities. All performance figures are company-reported.
Alibaba's Qwen Go! Adds My Apps, Building Shareable Apps From a Text Description via Meoo
Alibaba's personal agent Qwen Go! launched My Apps on October 23, letting users describe an app and have it built, configured and deployed without code. The feature runs on Alibaba Cloud's Meoo platform and is in invite-only testing.
Tencent publishes Hy-Nomas in Diffusers format: 4-step distilled video world model at 704 × 1280, 24 FPS
Tencent has posted a BF16 Diffusers version of the distilled Hy-Nomas video world generation model on Hugging Face. It generates one video segment at a time from text and camera input, using 4 denoising steps at 704 × 1280 and 24 FPS in the example configuration. Tencent also releases Hy-Nomas-Bench, a 280-case benchmark suite.
Comments
Loading...