Model

Gemini 2.5 Flash Preview (2025-09)

Google

deepmind.google

Agentic / Tool Use

Agentic / Tool Use
BenchmarkScoreMetricHarnessSourceConfidenceLink
Terminal-Bench17.1accuracyMini-SWE-AgentIndependentmediumtbench.ai
Terminal-Bench16.9accuracyTerminus 2Independentmediumtbench.ai
Terminal-Bench16.4accuracyOpenHandsIndependentmediumtbench.ai
Terminal-Bench15.4accuracyGemini CLIIndependentmediumtbench.ai
TheAgentCompany41.1resolved_rateN/AIndependentmediumthe-agent-company.com

Reproduced by the community

Loading community runs…