DeepSeek 深度求索

Only confirmed, sourced steps count. Announcements without follow-through are listed as declared interest at most.

• No direct European footprint or operations.Europe is currently a consumer of that experiment rather than a participant in it.Open Weights, Two Meanings: What China’s Four-Model Month Reveals
• European companies are running DeepSeek models on their own infrastructure.any enterprise can download the weights and run them on its own infrastructure — which is precisely what companies from Shenzhen to Stuttgart are quietly doing.China’s AI Just Raised Prices — and It’s Still Winning the Price War
• European customers/operationsSeveral European enterprises have already begun deploying self-hosted DeepSeek-R1 instances.DeepSeek: The Chinese AI Lab That Shocked Silicon Valley
• European regulatory attentionThe EU AI Office is tracking DeepSeek under the AI Act’s General-Purpose AI provisions, given its systemic risk threshold.DeepSeek: The Chinese AI Lab That Shocked Silicon Valley
• European data protection investigationsItalian and Irish data protection authorities launched investigations into DeepSeek’s data practices in early 2025, with Italy briefly blocking the service.DeepSeek: The Chinese AI Lab That Shocked Silicon Valley

Each line links to the article where we checked it. Company claims are marked.

• DeepSeek released V4-Flash, a 304B parameter model, explicitly built for inference speed.DeepSeek V4-Flash — 304 billion parameters, by far the smallest, and explicitly built for inference speed. The card documents DSpark speculative decoding with a draft module shipped inside the checkpoint (model card).Open Weights, Two Meanings: What China’s Four-Model Month Reveals
• V4-Flash is released under the MIT licence.DeepSeek V4-Flash and GLM-5.3-Flash are both released under the MIT licence.Open Weights, Two Meanings: What China’s Four-Model Month Reveals
• V4-Flash scored 54.4 on DeepSWE, according to Moonshot’s measurement.company claima coding-agent benchmark designed to resist data contamination, against 54.4 for DeepSeek V4-Flash.Open Weights, Two Meanings: What China’s Four-Model Month Reveals
• DeepSeek raised API prices on August 16, 2024.On August 16, DeepSeek raised its API prices, in some configurations by up to twelve times, introducing peak and off-peak billing.China’s AI Just Raised Prices — and It’s Still Winning the Price War
• DeepSeek’s V4-Pro model remains significantly cheaper than US competitors after the price hike.after that increase, its frontier model still costs five to twenty-five times less than comparable US APIs.China’s AI Just Raised Prices — and It’s Still Winning the Price War
• DeepSeek publishes its model weights.DeepSeek publishes its weights.China’s AI Just Raised Prices — and It’s Still Winning the Price War
• DeepSeek has adapted its frontier models to run on Huawei’s chips.DeepSeek has been adapting its frontier models to run on Huawei’s chips — serious AI performance without a single US GPU.AI Without Nvidia: China’s Silicon Declaration of Independence
• DeepSeek launched its V4-Pro model this week.When DeepSeek launched its V4-Pro model this week, the most important detail was not the benchmark scores.AI Without Nvidia: China’s Silicon Declaration of Independence