Android Developers Post New Android AI Benchmark Scores
Android Developers reportedly posted Android Bench scores for OpenAI GPT-6.1 Terra, Kimi K3.1 Pro, and DeepSeek V4.1 Flash on X on November 10. [[fact-jnxnldu52]] Android Bench is a Google AI programming benchmark for Android development that assesses real engineering tasks. [[fact-5a6jw4a73]] Kimi K3.1 Pro, using the kimi-code agent, reportedly recorded a 24.0% pass rate across 30 long-term tasks, 82.3% average completion, 27.4 hours average time, and $317.4 cost per run. [[fact-jveaqry2w]] The report says Kimi scored above GPT 6 Astra, Claude Fable 5.1, and Gemini 4 Argon, but below GPT 6.1 Terra at 28.7%. [[fact-sjwgla07p]] DeepSeek V4.1 Flash 1017, evaluated with the opencode agent, reportedly achieved a 17.3% pass rate, 71.2% average completion, 12.0 hours average time, and $9.9 cost per run. [[fact-8lxgbu4no]]