ResearchBenchmarksSoftware Engineering Problem Solving on SWE-bench Lite (test)Software Engineering Problem Solving on SWE-bench Lite (test)Follow57.3333Resolution RateCodex CLI56.98666757.07666757.16666757.256667May 7, 2026Evaluation resultsMethodMethodLinksResolution RateCost ($)Score (p)Total Tokens (M)Codex CLISystem=Codex CLISystem=Codex CLI2026.0557.3333161.821155.96SpellSystem=SpellSystem=Spell2026.0557102.12151.37