GPT-6 Sol leaks, screenshots and release rumors

The author fed a muted video to Codex (GPT-6 Astra/Sol) to design sound effects without a DAW or third-party samples, and shared a reusable prompt.

The author released a SKILL.md and GitHub repo showing how GPT-6 Astra (or Sol) and Opus 5.5 turn real estate photos into 3D walkable spaces.

A short post claims that simply telling GPT-6 Sol in xhigh mode to "make a breakthrough" really does work.

A user reports that GPT-6 Sol High misses instructions and rewrites text into a "neutral" tone even when it isn't asked to, calling it a "slop factory."

The author shares a workflow using Gemini 3.8 Flash TTS for voice and Wan for video generation, with Astra as the main tool and GPT-6 Sol as support.

A video shows GPT-6 Sol on Combos Fun iterating faster with fewer tokens, and the platform offers a Sol prompt collection to help users start building games.

The author used GPT-6 Sol with Combos CLI to build a playable online 12-level card game, with in-game images generated by Sol calling GPT-image 2.5, and shared a prompt collection.

The author says switching Codex to GPT-6 Sol cut development time by 90% and token use by half with no Skills installed, and shares a minesweeper prompt plus a demo of the finished game.

A full Japanese prompt for turning a full-body photo into a fisheye-lens subway car ad poster, which the author says should also work on 5.6 Sol.

elder_plinius released the full system prompts and tool definitions for GPT-6 Sol Codex, over 294,000 characters of extracted text, with a GitHub link.

A Japanese post collects an overseas user's method for dividing work among the three models: Astra for multi-step jobs, Sol for complex reasoning and agent work, Luna for repetitive extraction, plus token-saving trade-offs.

The post says changing reasoning effort mid-conversation does not invalidate the prompt cache for GPT-6 Astra, and OpenAI docs imply the same for upcoming families like Sol and Luna.