Google Unveils Gemini 3.6 Flash Models, Prepares for Gemini 4

WHO:

MarkTechPost - Google is the organization releasing the new Gemini models.

TechCrunch - The article discusses Google's AI strategy and the absence of the 3.5 Pro model.

WHAT:

VentureBeat - Gemini 3.6 Flash and Gemini 3.5 Flash-Lite are available immediately through the Gemini API in Google AI Studio and Android Studio, as well as within the consumer Gemini application and Google Search.

VentureBeat - Gemini 3.5 Flash Cyber will be available "exclusively to governments and trusted partners via CodeMender soon."

VentureBeat - Gemini 3.6 Flash features a 1-million-token input context window alongside a max output limit of 64,000 tokens, with both models sharing a knowledge cutoff date of March 2026.

VentureBeat - Gemini 3.6 Flash is priced at $1.50 per one million input tokens and $7.50 per one million output tokens through its API.

VentureBeat - Gemini 3.5 Flash-Lite costs $0.30 per million input tokens and $2.50 per million output tokens.

VentureBeat - Gemini 3.5 Flash Cyber will be available "exclusively to governments and trusted partners via CodeMender soon."

VentureBeat - Gemini 3.6 Flash scores 49% on the DeepSWE benchmark, an increase from 37% achieved by version 3.5.

VentureBeat - Gemini 3.6 Flash scores 63.9% on MLE-Bench compared to 49.7% previously.

VentureBeat - Gemini 3.5 Flash Cyber is designed specifically to find and fix cybersecurity vulnerabilities and integrates with Google’s CodeMender agent.

MarkTechPost - Flash-Lite runs at 350 tokens/sec.

MarkTechPost - Flash Cyber powers CodeMender for vulnerability finding.

Techmeme - Gemini 3.5 Flash Cyber is described as a "cost-efficient and highly capable alternative" to models like Mythos, available first to governments and some partners.

Techmeme - Gemini 3.5 Flash integrates into CodeMender, allowing it to identify and patch vulnerabilities.

MarkTechPost - Google released Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber.

MarkTechPost - The 3.6 Flash model reduces output tokens by 17% and its price is set at $7.50 per 1 million tokens.

Techmeme - Gemini 3.6 Flash improves coding, knowledge work, and multimodal performance.

Techmeme - Google has started its "most ambitious pre-training run yet" for Gemini 4.

WHERE:

Techmeme - The announcement pertains to Google's AI models, which are part of their product offerings.

WHEN:

MarkTechPost - The release of Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber occurred on July 21, 2026.

WHY:

Ars Technica - The release of the new models raises questions about Google's AI strategy, particularly regarding the absence of the 3.5 Pro model.

HOW:

VentureBeat - Gemini 3.6 Flash achieves significant efficiency gains by taking fewer reasoning steps and tool calls to complete multi-step workflows, resulting in reduced output token usage by 17% compared to its predecessor.

VentureBeat - Gemini 3.6 Flash achieves significant efficiency gains by taking fewer reasoning steps and tool calls to complete multi-step workflows.

Techmeme - The new models are designed to deliver efficiency, latency, and reliability for building AI agents at scale.

(UPDATED July 21, 2026 22:18 UTC)

Gemini 3.6 Flash and Gemini 3.5 Flash-Lite are now available through the Gemini API in Google AI Studio and Android Studio, as well as in the consumer Gemini application and Google Search. Additionally, Gemini 3.5 Flash Cyber will soon be available exclusively to governments and trusted partners via CodeMender. The new Gemini 3.6 Flash boasts a 1-million-token input context window and a maximum output limit of 64,000 tokens, achieving a 17% reduction in output token usage compared to the previous version by streamlining multi-step workflows.

(UPDATED July 21, 2026 20:22 UTC)

Gemini 3.6 Flash has been released with a price of $1.50 per one million input tokens and $7.50 per one million output tokens through its API, showing a notable improvement in performance with a score of 49% on the DeepSWE benchmark, up from 37% with version 3.5. Additionally, Gemini 3.6 Flash scored 63.9% on MLE-Bench, a significant increase from the previous 49.7%. Meanwhile, Gemini 3.5 Flash Cyber, designed to identify and address cybersecurity vulnerabilities, will soon be available exclusively to governments and trusted partners via CodeMender.

(UPDATED July 21, 2026 19:24 UTC)

Flash-Lite is making waves by operating at an impressive 350 tokens per second, while Flash Cyber is enhancing CodeMender's capabilities for vulnerability detection. The Gemini 3.5 Flash Cyber model is being touted as a "cost-efficient and highly capable alternative" to existing models like Mythos, with initial availability for governments and select partners. Additionally, this new Gemini model integrates seamlessly into CodeMender, enabling it to effectively identify and patch vulnerabilities.

(UPDATED July 21, 2026 18:22 UTC)

Google has just released its new Gemini models, including Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber, with the 3.6 Flash model reducing output tokens by 17% and priced at $7.50 per million tokens. This launch comes as part of Google's ongoing AI strategy, which has sparked discussions about the notable absence of the 3.5 Pro model, while the company also embarks on its most ambitious pre-training run yet for Gemini 4. The new models are designed to enhance efficiency, latency, and reliability in building AI agents at scale, marking a significant step in Google's product offerings.