Poolside AI Introduces Laguna XS.2 and M.1: Agentic Coding Fashions Reaching 68.2% and 72.5% on SWE-bench Verified
Poolside AI launched the primary two fashions in its Laguna household: Laguna M.1 and Laguna XS.2. Alongside these, the corporate ...
Tag: SWEbench
Collectively AI Releases DeepSWE: A Absolutely Open-Supply RL-Educated Coding Agent Primarily based on Qwen3-32B and Achieves 59% on SWEBench
Collectively AI has launched DeepSWE, a state-of-the-art, totally open-sourced software program engineering agent that's educated completely by reinforcement studying (RL). ...
Increase Code Launched Increase SWE-bench Verified Agent: An Open-Supply Agent Combining Claude Sonnet 3.7 and OpenAI O1 to Excel in Advanced Software program Engineering Duties
AI brokers are more and more very important in serving to engineers effectively deal with complicated coding duties. Nonetheless, one ...
