# AI News Digest - 2026-07-22

1. Microsoft and Mistral struck a multi-billion-dollar deal to build AI infrastructure across Europe.

2. Google shipped three new Gemini Flash models, including the more efficient Gemini 3.6 Flash that used up to 65% fewer tokens and a cybersecurity model available only to governments and select partners, while its anticipated Gemini 3.5 Pro remained in training.

3. Alibaba unveiled Qwen-Image-3.0, an image generator that accepted prompts up to 4,500 tokens, rendered legible text as small as ten pixels, supported twelve languages natively, and produced complex layouts such as infographics in a single pass.

4. Moonshot's free open-source model Kimi prompted public disputes among current and former Trump administration AI advisers and raised questions about competitive pressure from Chinese models on US AI companies and related policy responses.

5. JudgeGPT was evaluated in a field experiment with 1,559 Pakistani judges and was found to boost case resolution by 6.3 percent, producing an estimated return of up to $38.50 per dollar invested, with benefits concentrated among judges who received hands-on training.

# References

1. [https://the-decoder.com/microsoft-and-mistral-strike-multi-billion-dollar-deal-to-build-ai-infrastructure-across-europe/](https://the-decoder.com/microsoft-and-mistral-strike-multi-billion-dollar-deal-to-build-ai-infrastructure-across-europe/)

[Microsoft and Mistral strike multi-billion-dollar deal to build AI infrastructure across Europe](https://the-decoder.com/microsoft-and-mistral-strike-multi-billion-dollar-deal-to-build-ai-infrastructure-across-europe/)

1. [https://the-decoder.com/google-ships-three-new-gemini-flash-models-but-its-frontier-3-5-pro-remains-lost-in-training/](https://the-decoder.com/google-ships-three-new-gemini-flash-models-but-its-frontier-3-5-pro-remains-lost-in-training/)

[Google ships three new Gemini Flash models but its frontier 3.5 Pro remains lost in training](https://the-decoder.com/google-ships-three-new-gemini-flash-models-but-its-frontier-3-5-pro-remains-lost-in-training/)

1. [https://the-decoder.com/alibabas-qwen-image-3-0-renders-full-infographic-grids-and-readable-ten-pixel-text-in-a-single-pass/](https://the-decoder.com/alibabas-qwen-image-3-0-renders-full-infographic-grids-and-readable-ten-pixel-text-in-a-single-pass/)

[Alibaba's Qwen-Image-3.0 renders full infographic grids and readable ten-pixel text in a single pass](https://the-decoder.com/alibabas-qwen-image-3-0-renders-full-infographic-grids-and-readable-ten-pixel-text-in-a-single-pass/)

1. [https://www.technologyreview.com/2026/07/20/1140675/chinas-ai-models-have-trumps-ai-world-at-war-with-itself/](https://www.technologyreview.com/2026/07/20/1140675/chinas-ai-models-have-trumps-ai-world-at-war-with-itself/)

[China’s AI models have Trump’s AI world at war with itself](https://www.technologyreview.com/2026/07/20/1140675/chinas-ai-models-have-trumps-ai-world-at-war-with-itself/)

1. [https://the-decoder.com/an-ai-system-helped-pakistani-judges-clear-massive-backlogs-at-38-50-return-per-dollar-invested/](https://the-decoder.com/an-ai-system-helped-pakistani-judges-clear-massive-backlogs-at-38-50-return-per-dollar-invested/)

[An AI system helped Pakistani judges clear massive backlogs at $38.50 return per dollar invested](https://the-decoder.com/an-ai-system-helped-pakistani-judges-clear-massive-backlogs-at-38-50-return-per-dollar-invested/)

For the site tree, see the [root Markdown](https://ixtj.dev/.md).