Google Unveils Three New Gemini Flash Models as Gemini 3.5 Pro Remains Delayed

Google announced three new Gemini Flash models—Gemini 3.6 Flash, Gemini 3.5 Flash‑Lite, and Gemini 3.5 Flash Cyber—while confirming that Gemini 3.5 Pro remains unreleased and Gemini 4 is already under development.

Gemini 3.6 Flash

Positioned as a general‑purpose model for production workloads, Gemini 3.6 Flash improves programming, knowledge‑work, and multimodal capabilities over Gemini 3.5 Flash. In the Artificial Analysis Index test it reduces output token consumption by 17% and requires fewer inference steps and tool calls in multi‑step workflows. Pricing is $1.50 per million input tokens and $7.50 per million output tokens, lower than the 3.5 version.

Benchmark results show notable gains: DeepSWE score rises from 37% to 49%; MLE Bench from 49.7% to 63.9%; OSWorld‑Verified from 78.4% to 83.0%; and GDPval‑AA v2 from 1349 to 1421, indicating higher precision and reduced code revisions.

Gemini 3.5 Flash‑Lite

Targeted at low‑latency, high‑throughput tasks such as agent search and document processing, Flash‑Lite is the fastest model in the Gemini 3.5 series, delivering 350 tokens per second in Artificial Analysis tests. Its cost is $0.30 per million input tokens and $2.50 per million output tokens, offering a strong price‑performance ratio compared with Gemini 3.1 Flash‑Lite.

Performance improvements include Terminal‑Bench 2.1 (31% → 54%), GDM‑MRCR v2 (60.1% → 72.2%), and GDPval‑AA v2 (642 → 1140). In several agent and programming benchmarks it surpasses Gemini 3 Flash, e.g., SWE‑Bench Pro 54.2% vs. 49.6% and OSWorld‑Verified 74.0% vs. 65.1%.

Gemini 3.5 Flash Cyber (CodeMender)

Designed for security‑focused workloads, Gemini 3.5 Flash Cyber builds on Gemini 3.5 Flash with micro‑tuning for vulnerability detection and remediation. It offers lower token cost than larger models and achieves near‑state‑of‑the‑art results on the CyberGym benchmark.

In the CodeMender system, multiple Flash Cyber agents collaborate to produce a consolidated security report. Google adopts a cautious rollout, initially offering the model to government agencies and trusted partners via limited pilot deployments to mitigate misuse risks.

Availability and Roadmap

Gemini 3.6 Flash and Gemini 3.5 Flash‑Lite are publicly available through Google AI Studio, Android Studio, and the Gemini Enterprise Agent Platform; the latter is also being integrated into Google Search. Google disclosed that Gemini 3.5 Pro is still in partner testing and will be released when ready, while Gemini 4 is already in the largest pre‑training effort to date.