Digital billboards across San Francisco are now showcasing a new chart that compares various artificial intelligence models based on their intelligence and cost efficiency. Fable 5 from AnthropicAI leads with a score of 60 but has a cost per task of $2.75, while Grok 4.5 from SpaceXAI, although scoring lower at 54, is significantly cheaper at $0.31 and also ranks on the Pareto frontier. This frontier represents models that achieve optimal performance in intelligence and cost, an important metric as cost efficiency becomes increasingly critical in evaluating model performance given the recent trend among major AI developers releasing models with enhanced capabilities.

SpaceX: SpaceX operates an AI division known as SpaceXAI that builds frontier models in the Grok family for tasks including coding and agentic work. In mid-July 2026, it released Grok 4.5 and began rolling out access across iOS, Android, web, and X platforms. The company integrates these AI efforts with its core operations and recent corporate developments.
Elon Musk: Elon Musk leads SpaceX and directs its AI initiatives through SpaceXAI. He has publicly promoted recent model releases such as Grok 4.5, emphasizing efficiency improvements. Musk shapes the strategic priorities for these AI projects in 2026.
Anthropic AI: Anthropic is an AI company developing advanced models such as the Fable series with a focus on capability and safety. In July 2026, it received approval from the Trump administration to restore public access to previously restricted models including Fable 5. The company continues to attract investment interest from major hardware players exploring AI integrations.

Model Releases: Major AI developers are launching new models with enhanced performance characteristics and broader platform availability.
AI Benchmarking: Cost efficiency alongside intelligence metrics is gaining prominence when comparing frontier AI models.