Devstral 2: Open-Weights Coding Model for Cline
Mistral AI's Devstral 2 hits 72.2% on SWE-bench Verified and integrates directly into Cline. Free during launch, then 7x cheaper than Claude Sonnet on real-world coding tasks.
cline
Updates from Cline AI coding agent
Cline v3.39: Explain Changes Cuts Code Review Friction
Cline v3.39 adds inline explanations for AI-generated diffs. Understand what changed and why without context switching between diff and chat.
Cline Adds Free "Microwave" Stealth Model for Coding
Cline's new free stealth model "microwave" is live. Built for agentic coding with 256k context, no usage limits during alpha. Access it now via the Cline provider.
DeepSeek V3.2 and V3.2-Speciale Now in Cline
DeepSeek V3.2 and V3.2-Speciale are now available in Cline. Both models integrate reasoning directly into tool execution for agentic workflows. V3.2 for daily coding; Speciale for hard problems.
Claude Opus 4.5 Now Available in Cline
Claude Opus 4.5 is now in Cline with 80.9% SWE-bench performance and up to 65% fewer tokens. $5/$15 per million tokens, 200K context.
Cline-Bench: Real-World Benchmark for Agentic Coding
Cline launches cline-bench, an open source benchmark built from real engineering failures where frontier models break. Backed by OpenAI, Mistral, and Nous Research, with $1M for open source contributors.
Cline 3.38.0: Gemini 3 Pro & Voice Coding
Cline 3.38.0 adds Gemini 3 Pro Preview for stronger reasoning and AquaVoice Avalon for voice coding that understands technical terms. Plus critical AWS Bedrock fixes.