Google DeepMind

Gemini 2.0 Flash

• Released 15 April 2025
• Framework: Frontier Safety Framework (FSF)
• 5 tests by the provider
• 1 source document

The provider's own tests

• content policy evaluation · The provider · Harmful content

  Reported result: lower violation rates in most modalities than gemini 1.5 pro, small regression on image-to-text

  For content policies, we see the Gemini 2.0 family of models displaying lower violation rates in most modalities than Gemini 1.5 Pro, which in turn was a significant improvement on Gemini 1.0.

  Google DeepMind · 15 April 2025 · Read the source
• image to text safety · The provider · Harmful content
  violation rate change vs gemini 1 5 pro 002: 1.5% change · Test design: lower numbers are the intended result

  Image to Text Safety Automated content safety evaluation measuring safety policies +1.5%

  Google DeepMind · 15 April 2025 · Read the source
• multilingual safety · The provider · Harmful content
  violation rate change vs gemini 1 5 pro 002: -1% change · Test design: lower numbers are the intended result

  Multilingual Safety Automated safety policy evaluation across multiple languages -1.0%

  Google DeepMind · 15 April 2025 · Read the source
• text to text safety · The provider · Harmful content
  violation rate change vs gemini 1 5 pro 002: -1% change · Test design: lower numbers are the intended result

  Text to Text Safety Automated content safety evaluation measuring safety policies -1.0%

  Google DeepMind · 15 April 2025 · Read the source
• tone · The provider · Misuse & jailbreak
  tone improvement vs gemini 1 5 pro 002: 1.5% change

  Tone Automated evaluation measuring objective tone of model refusal +1.5%

  Google DeepMind · 15 April 2025 · Read the source

No document we read states: whether a capability level was crossed, an independent or government test, a safeguard. No incident on record names this version.

Compare with other versions