DoubleChecks

What was claimed

Today we’re releasing GDPx, a benchmark for measuring agentic capabilities... even the best model scores just 12% (1/7)

Our verdict

Accurate

SemiAnalysis describes GDPx as a benchmark for measuring general agentic capabilities across valuable frontier domains.

Combined assessment13 sourcesChecked Nov 2, 2026

Why this verdict

  • The 12% result is 1/7.

    Needs context
    One assessment reached this conclusion

  • The notation (1/7) represents the best model's performance

    Supported
    One assessment reached this conclusion

  • Even the best model scores just 12%.

    Supported
    All assessments reached this conclusion

  • Today we’re releasing GDPx, a benchmark for measuring agentic capabilities... even the best model scores just 12% (1/7)

    Supported
    All assessments reached this conclusion

Evidence reviewed

13 sources

Perplexity, Claude and ChatGPT sources are combined here.

newsletter.semianalysis.comaiweekly.coaiweekly.codigg.combadinfluence.newstldr.techtldr.techarxiv.orgen.wikipedia.orgarxiv.org

Individual assessments

Assessment summary

The release of GDPx and its purpose are supported by the primary announcement.newsletter.semianalysis.com The 12% figure is also reported, but “1/7” is misleading because the benchmark reportedly covers 11 domains and 206 tasks, not seven.newsletter.semianalysis.com digg.com

Claims reviewed

The 12% result is 1/7.

Needs context
Available reporting says GDPx covers 11 domains and 206 tasks, with the score calculated under benchmark-specific scoring rules.aiweekly.co digg.com The sources do not support interpreting 12% as one successful task out of seven.

This claim may mislead readers without added context.

Even the best model scores just 12%.

Supported
The GDPx announcement states that the best model scores 12%, and reporting identifies Claude Opus 5.5 as the leader at 12.0%.newsletter.semianalysis.com digg.com

Today we’re releasing GDPx, a benchmark for measuring agentic capabilities.

Supported
SemiAnalysis describes GDPx as a benchmark for measuring general agentic capabilities across valuable frontier domains.newsletter.semianalysis.com

Assessment summary

  • The statement accurately reflects SemiAnalysis's official announcement of GDPx on November 2, 2026.
  • All factual elements are correct: GDPx is indeed a benchmark for measuring agentic capabilities, it was released on that date, and the best-performing model (Claude Opus 5.5) scored 12% overall.
  • The notation "(1/7)" refers to passing 1 out of 7 tasks in a specific domain context.

Claims reviewed

The notation (1/7) represents the best model's performance

Supported
The (1/7) notation refers to passing 1 out of 7 tasks in a specific domain, consistent with how GDPx reports results across its 11 domains newsletter.semianalysis.com.

GDPx is a benchmark for measuring agentic capabilities

Supported
Multiple sources confirm GDPx is designed to measure agentic capabilities across 11 high-value domains newsletter.semianalysis.com fourweekmba.com cryptobriefing.com.

The benchmark was released today (November 2, 2026)

Supported
SemiAnalysis released GDPx on November 2, 2026, as confirmed by multiple sources newsletter.semianalysis.com fourweekmba.com cryptobriefing.com.

Assessment summary

Claims reviewed

Today we’re releasing GDPx, a benchmark for measuring agentic capabilities... even the best model scores just 12% (1/7)

Supported
SemiAnalysis published a blog post announcing GDPx and reports the benchmark's top score is ~12%, which corresponds to the '1/7' phrasing. newsletter.semianalysis.com gdpx.semianalysis.com

Check your own claim

Paste any statement, headline, or AI answer. Independent assessments check it and compare sources.

Share this result