Google adds three Flash-tier Gemini models, targets agentic workloads
Tech & Launches ·
The new lineup pairs a general-purpose workhorse model with a stripped-down speed variant and a cybersecurity-specific model built for code-defense tooling.
Google has released Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber, extending the Flash line that began with 3.5 Flash, according to a company blog post. The release is aimed squarely at teams running production AI agents that need to cut latency and token spend without sacrificing output quality.
The flagship of the trio, 3.6 Flash, is pitched as an upgrade for coding, knowledge work, and multimodal tasks. Per the Artificial Analysis Index cited in the announcement, it cuts output token usage by 17% versus 3.5 Flash, with gains reaching as high as 65% on select benchmarks such as DeepSWE from Datacurve. Pricing is set at $1.50 per million input tokens and $7.50 per million output tokens, which Google says lowers the total cost of running multi-step agentic tasks. On other benchmarks, 3.6 Flash scored 49% versus 37% on DeepSWE, 63.9% versus 49.7% on MLE Bench, 83.0% versus 78.4% on OSWorld-Verified, and 1421 versus 1349 on GDPval-AA v2, each measured against 3.5 Flash.
Gemini 3.5 Flash-Lite is positioned as the fastest and cheapest model in the 3.5 class, generating 350 output tokens per second by the same Artificial Analysis Index measure, with Google saying it outperforms earlier Flash-Lite versions specifically in agentic workflows.
The third addition, 3.5 Flash Cyber, is a specialized model built for security use cases and is deployed alongside Google's CodeMender code-security agent, combining a narrowly focused model with agent infrastructure to handle orchestration demands unique to cybersecurity applications.
Beyond these three releases, Google notes that Gemini 3.5 Pro remains in partner testing ahead of a broader rollout, and that pre-training has begun on Gemini 4, described as the company's most ambitious training run to date. No timeline was given for either the 3.5 Pro release or Gemini 4's availability, and it remains unclear how the new Flash models will be priced or positioned relative to competing agent-oriented offerings once adoption data emerges.