FULL STORY

Gemini 3.6 Flash: From Rumors to Reviews

Following rumors and early hiccups, Google officially launched Gemini 3.6 Flash. Tests showed major speed and cost improvements, alongside the announcement of Gemini 4 pre-training.

2026-07-21 ~ 2026-07-27 · 5 episodes · 188 posts

Episode 1 · Google Reportedly Launching Agentic Gemini 3.6 Flash with Price Cuts (2026-07-21, 10 posts)

Recent leaks suggest that Google is gearing up to release a new generation of models in late July. With the highly anticipated Gemini 3.5 Pro still delayed, Google is reportedly replacing it with Gemini 3.6 Flash, a model specifically designed for agents and coding tasks. Leaked comparison tables and model cards indicate that this model offers improved performance at a lower price, sparking widespread attention and discussion within the AI community.

Key Details and Leaks

Rumors claim Google will release three new models at once: Gemini 3.6 Flash, 3.5 Flash-Lite, and the cybersecurity-focused 3.5 Flash Cyber, skipping the 3.5 Pro. Among them, Gemini 3.6 Flash, identified as gemini-3.6-flash-tiered, briefly appeared in Antigravity. Designed for complex knowledge work, coding, and agentic workflows, the model reduces output tokens by 17% and jumps from 37% to 49% on the DeepSWE benchmark, featuring a lower price per million output tokens. Additionally, Reddit rumors suggest Google may have renamed or adjusted Gemini 3.5 Pro into the 3.6 Flash tier, with related internal model ID screenshots showing identifiers like flashLite and flash.

Community Reactions and Controversies

Community members are divided on this product route adjustment. @haider1 pointed out that while the current 3.5 Flash has strong agentic capabilities, its instruction-following is poor, often missing key details. @scaling01 joked that Google seems very cautious about the "Pro" route, suggesting that if the Flash version fails, they can still use a larger model as a backup. @Angaisb hopes the new model will be more token-efficient. @CtrlAltDwayne complained that this might just be "another expensive wrapper" product. Furthermore, @ChrisGPT joked that if OpenAI wants to disrupt Google's release rhythm, they should drop a "GPT 5.6" right now. Currently, there are no further parameters or official explanations regarding the new model.

Episode 2 · Gemini 3.6 Flash Glitch: Misidentifies Google's Latest Model (2026-07-21, 2 posts)

Google's newly released Gemini 3.6 Flash faced early mockery after it incorrectly identified Gemini 2.0 as the tech giant's latest model, despite having a May 2026 knowledge cutoff. Netizens quickly created memes mocking the AI blunder.

Episode 3 · Google Launches Three New Gemini Models, Announces Gemini 4 Pre-training (2026-07-21, 157 posts)

Google officially released three new models: Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber. They are now available on platforms like Google AI Studio, Vertex API, and Gemini App. Based on developer feedback, this update aims to deliver better large-scale agentic applications with higher intelligence, better token efficiency, and lower costs. Additionally, the company revealed that Gemini 4 has entered pre-training, and Elon Musk responded to the launch.

Confirmed

Gemini 3.6 Flash is positioned by the official team as one of the most capable models, focusing on delivering higher quality results with fewer tokens at the same cost. Compared to 3.5 Flash, it is priced lower ($1.50 per million input tokens, $7.00 per million output tokens) and reduces output tokens by up to 17% in complex workflows. It also shows improvements in writing production-level code, chart analysis, and document understanding. Meanwhile, 3.5 Flash-Lite is one of the smallest and fastest models, with an output speed approaching 350 tokens/s, priced the same as the retiring 2.5 Flash. Google also introduced 3.5 Flash Cyber, targeting cybersecurity to compete with Anthropic's Mythos, which is currently limited to government and trusted partners.

Unconfirmed

There is significant divergence regarding benchmark performance and actual cost-effectiveness. Developers like @CounterReady4774 showed strong agentic performance, and @bindureddy called it the "best chat model in the world." However, @bindureddy also noted that its benchmark scores are lower than the previous 3.5 Flash in some tests, and it is more expensive than Grok and Luna. @truecakesnake mentioned its scores are on par with 3.5 Flash based on Artificial Analysis data, and @XFreeze pointed out that its cost per task is higher than Grok 4.5. These controversies suggest the actual ROI might not fully meet official expectations.

Why it matters

Alongside the dense release of new models, Google's Logan Kilpatrick revealed that Gemini 4 has entered its "most ambitious" pre-training phase. According to @kimmonismus, Gemini 4 will be an entirely new foundation model, and the progress so far is exciting. This indicates Google is aggressively pushing to capture the AI market through rapid iteration and specialized models.

137 more related posts →

Episode 4 · Gemini 3.6 Flash Review: Faster and Cheaper, But Not Smarter (2026-07-22, 16 posts)

Google's latest release, Gemini 3.6 Flash, has sparked widespread testing and discussion within the community. The model achieves significant efficiency improvements, operating about twice as fast, reducing prices by 18%, and consuming fewer tokens. However, independent reviews generally note that its core intelligence has not been upgraded; its overall performance remains on par with the previous generation, with regressions in some capabilities, raising questions about its cost-effectiveness and upgrade value.

Confirmed

In terms of overall intelligence assessment, multiple authors point out that Gemini 3.6 Flash scores 50 on the Artificial Analysis (AA) Index, exactly matching its predecessor. @haider1 and @emax emphasize that although the new model shows progress in coding benchmarks (e.g., DeepSWE score increased from 37% to 49%), its overall performance still lags behind competitors like GPT-5.6 Sol and Terra, and it is criticized for being 2.5 times more expensive than GPT-5.6 Luna. Tests by @skalskip92 and @llamaindex reveal noticeable regressions in object detection and chart understanding (ParseBench dropped 14%), and @scaling01 notes that it often times out in WeirdML tests due to overly complex planning. In specific application scenarios, the model demonstrates strong capabilities: @allenainie mentions it achieved a high score of 68% in the browseruse web agent task test, surpassing GPT-5.6-sol and Sonnet 4.6; @mertdumenci highly praises its implementation experience in search-oriented tasks; and @bytebot confirms its multimodal capabilities remain solid and its web interface is extremely fast.

Unconfirmed

There is divided community feedback on whether the model feels smarter in subjective experience. @sankineth felt the model was "significantly smarter" after a brief trial, but this individual subjective experience differs from the "intelligence plateau" conclusion drawn from most benchmark tests.

Why it matters

The release strategy of this model reflects an adjustment in Google's roadmap. @PaiDxng believes Google is attempting to drive user upgrades through better official benchmark numbers and token optimization, but regression test results might hinder this process. A third-party report shared by @sujingshen indicates that Google is shifting its strategy towards reducing costs by 30-40% while maintaining capability parity. However, @burkov bluntly states that, based on data from arenas like the Frontend Code Arena, Google has now been surpassed by at least 6 other labs in the top-tier model competition. Overall, Gemini 3.6 Flash is an iteration focused on efficiency and specific agent capabilities rather than a leap in foundational intelligence.

Episode 5 · Google Unveils Three New Gemini Models Focused on Efficiency (2026-07-27, 3 posts)

Google has released three new Gemini models, including the Gemini 3.6 Flash, which is praised for delivering Sonnet-like quality at a significantly lower cost.