Liang Sheng flips the table again! DeepSeek's new model slays GPT-6, the era of wish-driven programming has arrived.

What happened: DeepSeek released V4.1-Flash-1017, an update to V4.1-Flash, across its web, App and API, with weights open on HuggingFace and ModelScope. It also opened its Search and Fetch APIs to the public. Key numbers: Search and Fetch APIs priced at 5 yuan per 1,000 calls. Why it matters: DeepSeek says the model leads Kimi K3, GLM 5.3, GPT 5.6-Sol and Claude Opus 5 on four benchmarks, and developers described it as fast and cheap. The update follows a nearly 5-hour API outage on October 15, after which DeepSeek apologized and refunded usage from October 8 to 14 as credit.

DeepSeek's new model outperforming GPT-6 signals major AI competitive shift.

DeepSeek's new model outperforming GPT-6 signals major AI competitive shift. — impact context not separately stored; see What changed.