OxCode benchmarks: 82% on two coding evaluations
OxCode benchmarks: 82% on SWE-bench Verified (410 of 500) and Terminal-Bench 2.1 (73 of 89), one attempt each. Read the method, caveats and run records.
Read articleFrom the team
Engineering, ideas, and the work behind the build.
1-5 of 5 articles
OxCode benchmarks: 82% on SWE-bench Verified (410 of 500) and Terminal-Bench 2.1 (73 of 89), one attempt each. Read the method, caveats and run records.
Read articleA step-by-step checklist for using an AI coding assistant for students in India on a real project, with code, expected output and checks you run yourself.
Read articleAn AI coding agent that runs tests should leave evidence. Use this checklist to verify written tests, execution, exit status, failures and reruns in your terminal.
Read articleTest a free AI coding agent for VS Code without risking your repo. Use this 5-step verification routine with throwaway branches and local tests.
Read articleCompare AI coding assistant pricing in rupees. See how OxCode plans, daily run limits, and sandbox verification protect your budget and billable hours.
Read article