Anthropic's recent releases of Claude Fable 5.1 and Opus 5 have driven the 71.5% market-implied odds, as these models currently lead independent composites on coding, reasoning, and agentic benchmarks such as SWE-Bench variants and Terminal-Bench, outperforming OpenAI's GPT-6 Astra and Google's Gemini 3.x series in several key evaluations. Traders appear to factor in Anthropic's emphasis on reliable tool use and large-context performance, alongside its internal deployment of models for roughly a quarter of R&D work, which signals accelerating capability gains through mid-2026. Google and OpenAI trail at 12.5% and 9.5% respectively due to slightly lower or less consistent scores on the hardest agentic tasks, while smaller labs like xAI and Meta sit under 3% amid limited frontier benchmark presence. Market sentiment reflects skin-in-the-game consensus on current trajectories, though model timelines and evaluation shifts could still alter the outcome by year-end.