GPT-6 Sol leaks, screenshots and release rumors

PawelHuryn shares bug-hunting benchmark data: GPT-6 Sol at max effort only matches GPT-5.6 Sol at medium and trails Opus 5.5 at medium.

AI coding tool TRAE says GPT-6 Sol is now available, aimed at complex coding and agentic tasks, with a screenshot of the interface.

A French post rounds up pricing, Artificial Analysis intelligence scores and real per-task cost, citing Dan Shipper's tests: Sol wins on writing, Opus 5.5 is stronger on long-running code.

A user argues Sol and Luna may not be a big jump over the previous generation, but lower prices and more usage matter more now that models are smart enough for most tasks.

A post with an image says Opus 5.5 and GPT-6 Sol made AI affordable again, claiming Sol is 79% cheaper than GPT-6 Astra and runs reliably on subscriptions.

A user shares and jokes about a video clip produced by GPT-6 Sol High, showing the model's output.

A video post says OpenAI released the Sol and Luna models, with API pricing of $2/$10 for Sol and $0.10/$0.50 for Luna, halving costs.

A user says 6-Sol costs a fifth of 5.6-Sol, and 6-Luna makes the already low 5.6-Luna price look expensive, calling both OpenAI and Anthropic winners.

A video pits Claude Opus 5.5, GPT-Astra 6 and GPT-6 Sol against each other on the same task to see which performs better.

A Japanese studio says it ran extensive validation tests over the holidays using Opus 5.5 plus GPT-6 Sol, and keeps posting Spirula Studio test results.

A Japanese poster's nightly news roundup says developers found GPT-6 Sol/Luna slightly behind Opus 5.5 in performance but far cheaper and more practical.

A list image counts seven releases in two days, including GPT-6 Sol and Luna, and says Grok 4.7 landed flat while Opus 5.5 took all the hype.