A new AI model, Sonnet 5, has achieved a score of 63.2% on the SWE-bench Pro benchmark. Separately, OpenAI's model, Terra, reportedly scored 84.3% on the Terminal-Bench, though this claim is vendor-stated, preview-only, and unverified. The author cautions against treating unconfirmed benchmarks as evidence, likening them to marketing press releases. AI
IMPACT Sets new benchmarks for AI model performance in software engineering tasks, potentially driving further development and competition.
RANK_REASON The cluster reports benchmark scores for AI models, which falls under research. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Mastodon — fosstodon.org →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
