VQV Signal

AI AT WORK WATCH TECHNICAL

Claude Sonnet 4.6 and GPT-6 Astra score 98% vs 99% on a SharePoint upgrade task

According to a post on X by @waldekm, Waldek Mastykarz compared two AI coding models on his team's SharePoint Framework upgrade task. Claude Sonnet 4.6 scored 98% at $0.92 a run, while GPT-6 Astra scored 99% at $5.14 a run. Both models were given the same skill and each was run five times. He urged developers choosing a default model for their coding agents to measure cost as well as score.

Source: Scobleizer News Pulse (X) · x.com Found via Robert Scoble’s Scobleizer News Pulse · a post on X, not independently verified Published 2026-11-20T14:53:00+00:00 Detected 2026-11-20T23:03:43+00:00
View original source

According to a post on X by @waldekm, Waldek Mastykarz compared two AI coding models on his team's SharePoint Framework upgrade task. Claude Sonnet 4.6 scored 98% at $0.92 a run, while GPT-6 Astra scored 99% at $5.14 a run. Both models were given the same skill and each was run five times. He urged developers choosing a default model for their coding agents to measure cost as well as score.

AI-assisted summary based on the listed source.

Waldek Mastykarz reported that Claude Sonnet 4.6 scored 98% on his team's SharePoint Framework upgrade task at $0.92 a run, while GPT-6 Astra scored 99% at $5.14 a run, with both given the same skill and run five times each. He urged anyone choosing the default model for their developers' coding agents to measure...

Teams choosing a default model for coding agents may see that a one-point accuracy gain can come with more than five times the cost per run. The post describes a single internal task, so the result may not carry over to other codebases or workloads. Readers should watch whether other teams publish similar comparisons that weigh cost alongside quality.

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

Signal Strength 91% Technical label WATCH Public Interest 30 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 28 Novelty Interest Score 70 Consequence Score 30 Curiosity Score 16 Shareability Score 46

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