Upgrading agentic coding capabilities with the new Devstral models

Selection record

AdmittedSum of both 143 ≥ twice the threshold 120

Source tier

Official, first-hand; this tier's threshold is 60

Pre-filter

passed：发布Devstral代码智能体模型及评测

Why it was chosen

SWE-Bench Verified scores, open-source licensing and API pricing for the two jointly released coding agent models enable cost and performance comparisons between open and closed approaches.

A model scores each item twice, independently, against one written standard, out of 100. An item is admitted only when the two scores add up to twice the threshold. The threshold is set per source tier.