AnalysisAI ModelsSeptember 28, 2026

GPT-6 Sol claims edge over Claude Opus 5 on business tasks at 9% cost

OpenAI launched GPT-6 Sol on September 22, claiming 33.2% on Zapier's AutomationBench versus Claude Opus 5's 26.9%, at 9% of the per-task cost. The New Stack questions whether cheaper matters when results aren't consistent.

How this story unfolded

9 days · 1 report · 12 community posts · from Sep 23

Sep 23
- Theory: Sol is the new Terra and should be compared to Sonnet not Opus
Sep 24
Sep 25
Sep 26
Sep 28
- GPT-6 Sol vs. Claude Opus 5.5: Is cheaper important when results aren’t consistent?thenewstack.io
- Is Chat area actually dying?
Sep 29
- sol 6 is a massive downgrade
Sep 30
Oct 2
- "Do X", "X is done!", "Are you sure?", "Yes!", "Did you test it?", "Yes!", "It doesn't look done and nothing works.", "I may have overstated completion. It is 7% done."

GPT-6 Sol claims edge over Claude Opus 5 on business tasks at 9% cost

How this story unfolded

More stories today

Coding Agent Index adds safety refusal reporting

GLM 5.3 and GLM 5.3 Flash land in Cursor

Hugging Face Hub to let users send community policies to Microduck robots

MiniMax H3 video generation extended to 30 seconds

Researcher: LeWM's per-timestep SIGReg allows temporal collapse

Acemoglu questions whether AI boom can continue without worsening inequality

Developer builds first OpenAI ModRetro Chromatic game

開発者がM5StackでDots向け携帯デバイスを自作