Google Gemini 3.6 Flash, 3.5 Flash-Lite, Cyber models

Google Releases Three New Gemini Models, Delays 3.5 Pro Launch

6 verified5 unconfirmed

Google DeepMind released three new AI models on Tuesday, including a next-generation Flash model and two specialized versions, but did not include the anticipated Gemini 3.5 Pro. The new models are Gemini 3.6 Flash, which replaces the recently deprecated 3.5 Flash, along with 3.5 Flash-Lite and 3.5 Flash Cyber. The company emphasized efficiency and cost reduction across the lineup, with the 3.6 Flash model using fewer tokens than its predecessor. The 3.5 Flash Cyber model was fine-tuned for cybersecurity tasks and is being offered through a limited access program. Google stated that it continues testing Gemini 3.5 Pro with partners and hopes to release it soon, while also beginning work on a more ambitious Gemini 4. The releases come amid a rapid pace of AI model launches from competitors.

What’s verified

Google released three new Gemini models: 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber.

Gemini 3.5 Flash has been deprecated and replaced by 3.6 Flash.

Gemini 3.6 Flash uses approximately 17 percent fewer tokens than its predecessor, improving efficiency.

Gemini 3.5 Flash Cyber is a specialized model for cybersecurity and is available only through a limited access program.

Gemini 3.5 Pro was not included in this release and remains delayed; Google says it is still testing the model with partners.

The company has started its next pre-training run for Gemini 4.

Not yet confirmed

One source reported specific benchmark scores for 3.6 Flash, including 49 percent on the DeepSWE coding test and 83 percent on the OSWorld computer use test, as well as support for a computer use feature in the API.

One source provided specific API pricing: 3.6 Flash at $1.50 per 1 million input tokens and $7.50 per 1 million output tokens, and 3.5 Flash-Lite at $0.30 per 1 million input tokens and $2.50 per 1 million output tokens.

One source noted that Google faced internal delays with 3.5 Pro due to performance goals, citing a Bloomberg report.

One source quoted Google DeepMind product lead Logan Kilpatrick stating the company hopes to "land soon" with 3.5 Pro.

One source listed recent releases from competitors OpenAI and Anthropic.