Skip to content
VibeCodeBench

OpenAI · Sol

GPT-6.2 Sol

Out since

Runs
25 runs on 5 prompts
Where I use it
  • ChatGPT Work · no runs yet

    Effort: Light, level 1 of 5Effort: Medium, level 2 of 5Effort: High, level 3 of 5Effort: Extra high, level 4 of 5Effort: Max, level 5 of 5

    Multi-agent mode, not an effort level: Ultra

  • Codex · 25 runs

    Effort: Low, level 1 of 5Effort: Medium, level 2 of 5Effort: High, level 3 of 5Effort: Extra high, level 4 of 5Effort: Max, level 5 of 5

    Multi-agent mode, not an effort level: ultra

These runs were made by a Codex agent: one answer each, never run or fixed by the agent.

Facts checked .

Not verified yet (2)
  • OpenAI does not say which effort levels each plan can pick for GPT-6.2 (Max and Ultra by plan are not stated); the ladder shown is the general Work/Codex one
  • Business: learn.chatgpt.com lists it for Work, OpenAI's announcement does not
Build a Cell, Part by PartGPT-6.2 Sol · Max
Build a Cell, Part by Part by GPT-6.2 Sol: a still of the demo
works · auto-checked (Checked by an automated pass. Not hand-played yet.)
Click it to play here.Playing here. Done gives the page back.

Its runs

What GPT-6.2 Sol built.

Broken tries stay on the page too.

Effort ladder

More effort, better result?

Same prompt. Only the effort changes.

Other generations

The Sol line.

  1. GPT-6.2 Solthis page · 25 runs
  2. GPT-6.1 Sol · 25 runs