Google and OpenAI lead the second-best AI agent lab market at 45.4% and 41.0% implied probability, reflecting closely matched trader sentiment on their agent capabilities. Recent benchmark results and model launches drive this positioning: Google’s Gemini 4 Carbon, rolled out November 10 with strong long-horizon and cybersecurity agent performance on APEX-Agents, competes directly with OpenAI’s GPT-6 series, which tops some agentic leaderboards like Agents’ Last Exam. Anthropic’s Claude Opus 5.5 and Fable 5.2 releases emphasize multi-step reasoning and enterprise agent workflows but trail at 14.5%. Differentiators include Google’s swarm features and enterprise infrastructure versus OpenAI’s computer-use tools and developer ecosystem. With resolution at month-end, fresh benchmarks or releases could shift the narrow gap.