VQV Signal

AI AT WORK WATCH TECHNICAL

Gemini 4 Hydrogen's "xhigh" effort cost about 6x more than "high" in one test

According to a post on X by @simonw, Simon Willison ran his pelican test on Gemini 4 Hydrogen at two thinking effort levels. On high, the test cost 3.99 cents and used 5,311 output tokens. On xhigh, the same model cost 23.61 cents and used 31,477 output tokens, and Willison called the difference in output very significant.

Source: Scobleizer News Pulse (X) · x.com Found via Robert Scoble’s Scobleizer News Pulse · a post on X, not independently verified Published 2026-11-21T20:30:00+00:00 Detected 2026-11-22T10:04:05+00:00
View original source

According to a post on X by @simonw, Simon Willison ran his pelican test on Gemini 4 Hydrogen at two thinking effort levels. On high, the test cost 3.99 cents and used 5,311 output tokens. On xhigh, the same model cost 23.61 cents and used 31,477 output tokens, and Willison called the difference in output very significant.

AI-assisted summary based on the listed source.

In his pelican test, Gemini 4 Hydrogen on high thinking effort cost 3.99 cents and used 5,311 output tokens, while the same model on xhigh cost 23.61 cents and used 31,477 output tokens; he called the difference in output very significant. Why it matters: Developers choosing Gemini 4 effort levels face a large...

Developers choosing effort levels for Gemini 4 Hydrogen may face a large trade-off between cost and output on the same model. Teams budgeting for API use would need to weigh whether the extra output at xhigh justifies a price about six times higher for their own workloads. This is a single test described in one post, so broader cost patterns would need more data.

Teams using AI at work may want to compare this against current productivity and review workflows.

Signal Strength 78% Technical label WATCH Public Interest 26 Category AI AT WORK Reader Depth TECHNICAL

Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.

Public Interest components
Recognizable Entity Score 0 Practical Impact Score 20 Novelty Interest Score 70 Consequence Score 22 Curiosity Score 0 Shareability Score 45

VQV surfaced this signal because it is recent, relevant to AI Chips, connected to Scobleizer News Pulse (X).