National Supercomputing Internet Cuts AI Model Calling Costs

What happened

National Supercomputing Internet announced price changes, including a limited-time offer for new users to call DeepSeek-V4.1-Flash-1017 and GLM-5.3-Flash for 0.1 yuan per 50 million tokens. It also added more mainstream models to its TokenPlan plans. Subscribers can call the Auto model for free from 21:00 to 09:00 nightly, October 21–30.

Read next

• More changes in today’s briefingConfigure the ChatGPT desktop app with Amazon Bedrock in AWS GovCloud (US)

The story so far

DeepSeek released DeepSeek-V4.1-Flash-1017 on its website, app, and API, with the update also available on Ollama Cloud and through a limited-time National Supercomputing Internet offer; Cursor has added DeepSeek V4.1 Flash. DeepSeek says the 1017 update improves benchmark performance, completes tasks faster, and produces shorter outputs than the previous version. It says the model is optimized for DeepSeek Harness, Tencent CodeBuddy, and WorkBuddy, and recommends DeepSeek Harness v0.2.1-rc.1 or later. Ollama says its deployment uses US-based hosting with zero data retention and supports access through Claude Code and OpenCode integrations under the tag deepseek-v4.1-flash:1017-cloud. Cursor users can enable DeepSeek V4.1 Flash in Settings > Models. Cursor says every effort level lies on its CursorBench cost frontier, with no model scoring higher at corresponding costs of $0.60–$0.98 per task. These are benchmark comparison figures. Cursor’s announcement names V4.1 Flash without specifying the 1017 update. Leiphone reports that National Supercomputing Internet is offering new users a limited-time price of 0.1 yuan per 50 million tokens for DeepSeek-V4.1-Flash-1017 and GLM-5.3-Flash.