Google DeepMind

Gemini 3.1 Flash-Lite

• Framework: Frontier Safety Framework
• 1 statement about capability levels
• 1 independent or government test
• 6 tests by the provider
• 4 source documents

What the provider states

• Other · CCL (Critical Capability Level) · Not determined

  our assessments have shown that gemini 3.1 flash-lite is less capable than gemini 3.1 pro, therefore based on gemini 3.1 pro, we are confident that gemini 3.1 flash-lite is also unlikely to reach any ccls.

  Google DeepMind · 3 March 2026 · Read the source

Independent and government tests

• judge evaluation against approximate ground truth severity labels · Apollo Research, independent evaluator · Deception & sabotage

  for example, gemini 3.1 flash-lite skips severity grading to score task execution, haiku 4.5 primarily fails to flag violations at all, and gpt-5.4 mini identifies violations but discounts them heavily through session-context framing.

  Apollo Research · 7 July 2026 · Read the source

The provider's own tests

• image to text safety · The provider · Harmful content

  image to text safety automated content safety evaluation measuring safety policies -21.7%

  Google DeepMind · 3 March 2026 · Read the source
• manual red teaming · The provider · Harmful content

  like 3 pro, the scope of red teaming covered potential issues outside of our strict policies, and found no egregious concerns.

  Google DeepMind · 3 March 2026 · Read the source
• multilingual safety · The provider · Harmful content

  multilingual safety automated safety policy evaluation across multiple languages -1.84%

  Google DeepMind · 3 March 2026 · Read the source
• text to text safety · The provider · Harmful content

  text to text safety automated content safety evaluation measuring safety policies -1.18%

  Google DeepMind · 3 March 2026 · Read the source
• tone · The provider · Misuse & jailbreak

  tone 2 automated evaluation measuring objective tone of model refusal +14.59%

  Google DeepMind · 3 March 2026 · Read the source
• unjustified-refusals · The provider · Misuse & jailbreak

  unjustified-refusals automated evaluation measuring model’s ability to respond to borderline prompts while remaining safe -14.41%

  Google DeepMind · 3 March 2026 · Read the source

No document we read states: a release date, a safeguard. No incident on record names this version.

Some documents we hold are stored in lower case only, so their quotes are shown in lower case. The source link opens the original.

Compare with other versions