Google Launches Gemini 3.6 Flash and Two Low-Cost AI Models

Google released three lightweight artificial intelligence models on Tuesday, lowering token costs for enterprise developers while releasing a restricted cybersecurity tool.

The new models, Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber, focus on processing speed and lower operational costs.

Gemini 3.6 Flash serves as the primary update for general coding and knowledge work.

Google cut output token consumption by 17% compared to Gemini 3.5 Flash on the Artificial Analysis Index, pricing the model at $1.50 per million input tokens and $7.50 per million output tokens.

The model achieved a 49% score on the DeepSWE software engineering benchmark and 83% on the OSWorld-Verified computer-use test.

Early corporate clients including Harvey AI and Hebbia integrated the model into legal document parsing and financial research workflows.

Niko Grupen, head of applied research at Harvey, evaluated the performance of the model during early testing.

"Gemini 3.6 Flash excels at document drafting and review in practice areas like capital markets and corporate M&A," Grupen said. "Compared to its predecessor, Gemini 3.6 Flash showed strong gains in performance on our benchmarks and was notably more efficient, completing tasks 12% faster on average."

Google simultaneously introduced Gemini 3.5 Flash-Lite for high-volume automated operations.

Generating up to 350 tokens per second at $0.30 per million input tokens and $2.50 per million output tokens, the model targets high-frequency tasks such as document extraction and web search queries.

It scored 54% on Terminal-Bench 2.1, up from 31% on the earlier 3.1 Flash-Lite model.

As Google claims, Gemini 3.5 Flash-Lite is the fastest and most cost-effective AI model from Google till now, and it just tops benchmarks too:

Chief Executive Officer Sundar Pichai previously highlighted corporate software budgets when discussing enterprise deployment costs.

"Companies are already blowing through their annual token budgets, and it's only May," Pichai said.

The third model, Gemini 3.5 Flash Cyber, operates inside Google's CodeMender security agent to detect and patch software code vulnerabilities.

During internal testing on the Chrome V8 JavaScript Engine, Flash Cyber identified 55 unique confirmed security flaws, surpassing rival security tools.

Google restricted distribution of Flash Cyber to government agencies and verified security partners to prevent potential misuse by cyber attackers.

Raluca Ada Popa, Gemini Security Lead at DeepMind, and Four Flynn, vice president of security and privacy at DeepMind, detailed the deployment strategy in a technical update.

"Given the dual-use nature of this technology, we have taken an intentional approach to how we deploy 3.5 Flash Cyber," Popa and Flynn said. "As part of a limited-access pilot program, 3.5 Flash Cyber will be exclusively available to governments and trusted partners via CodeMender, expanding over time. This will give frontline defenders a head start in finding and fixing critical vulnerabilities before they can be exploited, while mitigating against broader misuse."

Google held back its expected flagship model, Gemini 3.5 Pro, keeping it in private testing with selected partners following internal coding performance reviews.

Competitors Anthropic, OpenAI, xAI, and Moonshot recently launched flagship models, placing pressure on Google's model leaderboards.

Google confirmed that pre-training work has officially begun for its next-generation Gemini 4 release.