🤖 Googleന്റെ puthya gemini 3.6 flash

The Trump administration is hunting for "watermarks" in Chinese AI models, claiming they stole U.S. tech

Alibaba unveiled a 2.4 trillion parameter beast that undercuts GPT-5 pricing by 90%

Google drops three new Gemini models Flash gets faster, Lite gets cheaper

Trump administration targets open source AI for alleged IP theft from U.S. companies

Alibaba's Qwen3.8-Max claims runner-up spot to Fable 5 at a fraction of the price

Gemini 3.6 Flash is the new workhorse, costing $1.50 per million input tokens and scoring 49% on the DeepSWE coding test.

Gemini 3.5 Flash-Lite is built for pure speed, pumping out 350 tokens per second for just $0.30 per million input tokens.

Gemini 3.5 Flash Cyber is a locked-down model for finding software bugs, available only to governments and trusted partners.

Google also confirmed they have already started training the next-generation Gemini 4.

Developers need cheap, fast models to build AI agents that actually work.

These new Flash models give them exactly that. By cutting token usage and boosting speed, Google makes multi-step AI tasks affordable.

They are filling a huge gap in the market while everyone waits for their heavy-hitter models.

Google is playing a smart game here. Instead of rushing a buggy Pro model, they are flooding the zone with cheap, fast tools developers actually want.

If you build AI agents, the new Flash-Lite model is an absolute steal.

The US Treasury claims Chinese companies hit American AI systems hundreds of millions of times to copy their code.

This crackdown comes right after China's Moonshot AI launched Kimi K3, a model that beats OpenAI's GPT-5.6 Sol on price.

Bessent called these foreign models 'counterfeit goods' and warned that overseas labs will face strict economic sanctions.

The government might even force US businesses to declare if they are using any Chinese AI tools.

This is a massive jump in the tech war between the US and China.

AI models are now getting treated like stolen military secrets instead of open-source software.

If the US starts banning these models, it will split the global AI community in half.

Companies that rely on cheap open-source tools will suddenly face huge legal risks.

You cannot stuff the open-source genie back in the bottle.

Trying to ban developers from using cheap, powerful models like Kimi K3 is a losing battle.

The US needs to focus on building better AI instead of just policing the internet with sanctions.

Qwen3.8-Max packs 2.4 trillion parameters and handles text, images, video, and documents.

Alibaba claims the model ranks second only to Fable 5, but no outside groups have tested this yet.

The preview version went live on July 19, 2026, with prices slashed as low as $0.17 per million input tokens.

Developers can use the preview now on Alibaba Cloud, and the company promises a full open-weight release in the future.

The AI race in China is heating up fast.

Just days after Moonshot AI launched a massive model, Alibaba fired back with an incredibly cheap alternative.

By dropping prices to a fraction of normal costs, Alibaba wants to force a global price war.

This makes it much cheaper for you to build AI apps while putting heavy pressure on American companies to lower their own fees.

Claiming you are the second-best model in the world means nothing until independent testers actually prove it.

But honestly, at $0.17 per million tokens, developers will flock to it anyway just to save cash.

Alibaba is buying market share, and if the model is even half as good as they say, it is a massive win for builders.

Google's Flash lineup: Gemini 3.6 Flash and two 3.5 variants drop faster, cheaper, and built for coding and cybersecurity.

Trump targets open source: U.S. Treasury claims Chinese AI models stole American tech and plans to crack down on IP theft.

Alibaba goes cheap: Qwen3.8-Max launches with 2.4 trillion parameters and pricing so aggressive it undercuts GPT-4o by 95%.