LessWrongLessWrong · LessWrong published a summary of a viral Chinese article about a DeepSeek kernel engineer's views on automating his own work.
epoch.ai, Epoch AI · 3 sourcesepoch.ai · 3 sources · 4 reports · Epoch AI released the InnovationEval benchmark, testing whether frontier AI can independently discover machine learning algorithm innovations that are new to it, benchmarked against the Self-Distillation Policy Optimization (SDPO) proposed in a human paper. The task requires an agent to start from GRPO and develop new post-training improvement methods, improving short-answer and coding task performance on Qwen3-8B, with a budget of 3,000 GPU hours. Results show that GPT-5.6 Sol and Claude Fable 5 both did not come close to SDPO: Sol's method, adjusted for the same wall-clock time, reached only 15% of SDPO's gain, or about 35% under lenient scoring; Fable 5 showed no measurable improvement. Both merely reused known techniques. Epoch AI also estimates that, based on AI chips shipped before 2027, if cheap models are run, the world could support nearly 2 billion concurrent AI agents, while running frontier models would support about 30 million to 170 million; the cost of reaching a fixed AI performance level has fallen about 47% per quarter over the past three years.
CNBC TV18 / Bloomberg, The Decoder · 4 sourcesCNBC TV18 / Bloomberg · 4 sources · 4 reports · According to Bloomberg, citing sources familiar with the matter, DeepSeek's latest funding round is close to finalizing at least 80 billion yuan (about 12 billion USD), with Tencent Holdings and CATL committing the largest amounts. DeepSeek initially planned to raise about 50 billion yuan, but due to investor demand exceeding expectations, the final size could approach 100 billion yuan. Sources say that after the funding round is completed, the next step is to restructure the company to prepare for an IPO in early 2027, with an initial target valuation of about 500 billion yuan. CNBC separately reported that DeepSeek is considering expanding the funding round to up to 100 billion yuan, double the previous 50 billion yuan target; the round was initially expected to close by the end of August but was delayed because some investors hesitated on price, and negotiations are nearing completion. DeepSeek, Tencent, and CATL did not immediately respond to requests for comment.
