Latest
  1. Johns Hopkins Bayview tells 3,182 patients an employee put their health data into a public AI tool
  2. China’s AI compute access expected to double in 2027, with US-designed chips about 65 per cent
  3. UK commits £100m to national AI Biosecurity Lab
  4. Moonshot AI releases Kimi K3.1 Pro with 1-million-token context window
  5. Residents in Newport and Cardiff oppose hyperscale datacentres near homes
  6. Study: Grok 4.7 picked Republicans in 92% of forced midterm answers, GPT-6.1 Sol picked Democrats in 97%
  7. Attackers exploiting the PHP flaws hand the break-in to AI agents, Unit 42 says
  8. Elvis Presley’s rightsholders sue Anthropic over training on 274 songs

Latest / Frontier models and safety

Model releaseFrontierUSConfirmed

Google releases Gemini 3.1 Pro, doubling Gemini 3 Pro’s ARC-AGI-2 score

See how this story connects

Source: Google: Gemini 3.1 Pro

Google released Gemini 3.1 Pro across the Gemini API, Vertex AI, the Gemini app and NotebookLM, a week after a major Gemini 3 Deep Think update. Google said it scored a verified 77.1% on ARC-AGI-2, more than double Gemini 3 Pro.

Why it matters

Google’s mainline model made a large jump on a reasoning test built to resist memorisation.

When
19 Feb 2026
ID
WTA-N-2026-0157
Who
Google, Google DeepMind

Summary by wheresthe.ai, from Google. Checked against the source on 9 Oct 2026.

Read the original at Google

Source copy: download the page as saved on 18 Oct 2026 (PDF). Also held by the Internet Archive.

Next stop on this Storyline: Led toGoogle launches Gemini 3.5 with 3.5 Flash for agents and coding

Cite this story

WTA-N-2026-0157: “Google releases Gemini 3.1 Pro, doubling Gemini 3 Pro’s ARC-AGI-2 score.” wheresthe.ai, 19 Feb 2026. https://wheresthe.ai/d/google-gemini-3-1-pro-2026-02/ (source: Google: Gemini 3.1 Pro).

Spotted an error? Email [email protected].