OpenAI Extends GPT-6 Line with Cheaper, More Accurate Sol and Luna Models

Skynet Chance (+0.02%): Halving the cost of near-frontier intelligence broadens deployment of capable agentic models into coding and workplace pipelines, expanding the attack surface for misuse and unmonitored autonomous action even as factuality improves. The 90-minute competitive leapfrog with Anthropic also signals release cadence driven by market pressure rather than safety review.

Skynet Date (-1 days): Cheaper, more reliable models accelerate the diffusion of autonomous AI agents across real economic workflows, pulling forward the point at which significant capability is operating with limited human oversight. The tight competitive release cycle further compresses deliberation time between capability jumps.

AGI Progress (+0.02%): This is a distillation and efficiency milestone rather than a capability frontier push, but reaching flagship-level reliability at a fraction of the cost with a measurable halving of factual errors is meaningful progress on the robustness problem that limits general-purpose deployment.

AGI Date (+0 days): Dramatic cost reductions in inference mean more compute available for experimentation, agentic scaffolding, and synthetic data generation, which historically accelerates the overall development loop toward more general systems.