Gemini 3 Pro: corrected the stored Humanity's Last Exam figure to Google's published 37.5% no-tools launch result

The OnAirToday AI Progress Tracker recorded a verified change to Google's Gemini 3 Pro (version Gemini 3 Pro Preview): Corrected the stored Humanity's Last Exam figure to Google's published 37.5% no-tools launch result.

Gemini 3 Pro launched as Google's multimodal reasoning model in preview; the specifications and benchmark results below describe that launch, not a verified current deployment. Live source retrieval is unavailable here, so current availability, pricing and successor releases remain unverified, and the incident entries represent supplied reporting rather than independently checked findings.

What changed in this review:

• Added vendor launch results for GPQA Diamond, SWE-bench Verified, ARC-AGI-2 and Terminal-Bench 2.0.
• Set current prices to null because launch-era rates cannot establish September 2026 pricing.
• Added each supplied security story separately; entries appear to cover the same reported event, not sixteen distinct incidents.
• Gemini 3 Pro attribution is unconfirmed: the supplied incident coverage does not identify the Gemini version.
• Current lifecycle, successor releases and source reachability still require live verification; this is not a completed current-state recheck.
• Capability score 37 → 62

Anyone building on Gemini 3 Pro should re-check the model's risk dashboard before shipping, since flaws and reported incidents are what moved in this review.

The full profile — capabilities, benchmark citations, known flaws and risk guidance — is on the tracker at /ai-models/gemini-3-pro.

References:

• https://blog.google/products/gemini/gemini-3/
• https://ai.google.dev/gemini-api/docs/gemini-3
• https://ai.google.dev/gemini-api/docs/pricing
• https://ai.google.dev/gemini-api/docs/deprecations