GPT-6 Sol leaks, screenshots and release rumors

A post claims Opus 5.5 launched the same day as GPT-6 Sol with lower cost and stronger results on some benchmarks, and says Sonnet 5.5 and Haiku 5.5 are coming for Sol and Luna.

A Base ecosystem daily roundup covers on-chain news and closes with one line noting that usedotai launched anonymous versions of GPT-6 Sol and Luna.

A developer says GPT-5.6 Sol reached 35% of tests passing and GPT-6 Astra 85% before both stalled in loops, while Opus 5.5 completed the port in 10 hours.

A post says even the unreleased Claude Sonnet 5.5 beats GPT-6 Sol and questions how OpenAI will recover at DevDay.

A user claims Opus 5.5 beat GPT-6 Astra and GPT-6 Sol at modeling a Roman legionary in Blender, winning on textures, detail, accuracy, materials and realism.

A video compares games generated by Claude Opus 5.5 and GPT-6 Sol to see which model built the better one.

Under a 20-step or 5-minute limit, Opus 5.5 solved a Rubik's cube in three steps, GPT-6 Sol timed out after a wrong first move, and Jev and Laya used all 20 steps without solving it.

A joking post puts GPT-6 Sol, Opus 5.5 and Grok 4.7 into the classic fuck/marry/kill game, with no substantive information.

More than 10 models shipped in a week, including GPT-6 Sol, Luna, Opus 5.5 and Grok 4.7, and the author rates Opus 5.5 as the best release.

A 60-second video recaps the week in AI, led by the arrival of GPT-6 Sol and the release of Claude Opus 5.5.

A user says the 50% price cut on GPT-6 Sol also reduced quality, and that their earlier criticism of the model was not harsh enough.

On Parallel's search capability leaderboard, GPT-6 Luna is Pareto-optimal and leads on efficiency, while GPT-6 Sol matches GPT-5.6 at about 38% lower cost.