← sources

GPT-6 and Opus 5.5's biggest revolution isn't performance, its speed and cost.

source: reddit (reddit:singularity)published 2026-09-23 09:50ingested 2026-09-24 02:38

retained text

Low reasoning GPT-6 Sol and Opus 5.5 almost rival last generation's medium reasoning performance at 1/4th of the price while being 4x faster to complete the task.

Frontier agent performance levels at budget model prices and result speed. The singularity is here? Data from [artificialanalysis.ai](http://artificialanalysis.ai)

|Model (reasoning level)|Intelligence Index|Cost per Task|Time per Task|Total Output Tokens (full Index run)|
|:-|:-|:-|:-|:-|
|GPT-6 Sol (medium)|40|$0.25 |59.99s|16M|
|**GPT-6 Sol (low)**|34|$0.13 |27.75s |9M|
|Claude Opus 5.5 (medium, fallback)|51|$1.34|206.07s|38M|
|**Claude Opus 5.5 (low, fallback)**|42|$0.55|81.35s|20M|
|GPT-5.6 Sol (medium)|39|$0.50|125.03s|21M|
|Claude Opus 5 (medium)|45|$2.19|320.93s|49M|

document fetched 2026-09-24 00:33 · canonical https://reddit.com/r/singularity/comments/1wo1ij3/gpt6_and_opus_55s_biggest_revolution_isnt