AI modelsNovember 18, 2026
22
š¶ļø
GPT-6.2 Sol: Pricier, Not Better
A user report finds GPT-6.2 Sol leads the leaderboard yet matches prior quality on practical tasks while costing 25 percent more.
#OpenAI#LLM#benchmark#coding#cost

š„ What happened
OpenAI shipped GPT-6.2 Sol at the same price as 6.1 Sol. One engineer replayed five identical coding orders through both models: 6.2 Sol cost 25% more and delivered no quality gain.
š” Why it matters
A single outlier order decided the result: 6.2 Sol stopped before committing and triggered a costly retry. Strip that case out and 6.2 Sol is 13% cheaper. The Intelligence Index shows +7 points, but the real workflow tells the opposite story.
ā” Our take
Benchmarks lie, replays don't. Anyone swapping default models without replaying their own paid orders is burning money ā here, $11 to learn that 6.1 Sol stays.
The title, summary and analysis of this item were produced automatically by an AI system and have not been editorially reviewed. They may contain errors, bias or omissions ā when in doubt, read the linked original source.
Deep Dives & Similar Intelligence
ki-daily.
OpenAI slashes prices, dominates benchmarks
AI modelsNovember 17, 2026

LLM Traders: Only Memorization Beats the Market
ResearchNovember 17, 2026
Alibaba open-sources its code review agent
Tools & ProjectsNovember 13, 2026
OpenAI Takes Aim at GitHub Security
Tools & ProjectsNovember 12, 2026

165M Model Trained for $4,200
ResearchNovember 15, 2026