This gain is made possible by TNG’s Assembly-of-Experts (AoE) method — a technique for building LLMs by selectively merging the weight tensors ...

German firm TNG has released DeepSeek-TNG R1T2 Chimera, an open-source variant twice as fast as its parent model thanks to a ...

The Hangzhou-based firm said later in a short post on X that R1-0528 featured improved performance. In a longer post on WeChat, DeepSeek said the rate of "hallucinations", false or misleading ...

Smaller Variants for Scalable Deployments. For enterprises with limited compute resources, DeepSeek has introduced a distilled version, DeepSeek-R1-0528-Qwen3-8B, optimized for smaller-scale ...

DeepSeek also said it distilled the reasoning steps used in R1-0528 into Alibaba’s Qwen3 8B Base model. That process created a new, smaller model that surpassed Qwen3’s performance by more ...

Deepseek R1-0528 challenges proprietary AI models like OpenAI’s GPT-4 and Google’s Gemini 2.5 Pro by offering comparable performance at significantly lower costs, providing widespread access ...

Deepseek’s R1-0528 AI model competes with industry leaders like GPT-4 and Google’s Gemini 2.5 Pro, excelling in reasoning, cost efficiency, and technical innovation despite a modest $6 million ...

DeepSeek released an updated version of their popular R1 reasoning model (version 0528) with – according to the company – increased benchmark performance, reduced hallucinations, and native support ...

DeepSeek's updated R1 AI model is more censored than the AI lab's previously releases, one test found — in particular when it comes to criticism of the Chinese government.

The DeepSeek-R1-0528 model brings substantial advancements in reasoning capabilities, achieving notable benchmark improvements such as AIME 2025 accuracy rising from 70% to 87.5% and LiveCodeBench ...

Why DeepSeek-R1-0528 Matters for GPTBots.ai Users. ... This variant achieves state-of-the-art performance among open-source models while requiring only 16 GB of GPU memory, ...