DeepSeek Updates V4.1 Flash and Opens Search APIs

DeepSeek announced an update to its V4.1 Flash artificial intelligence model on October 17, saying it improves benchmark results while completing tasks faster and producing shorter responses. The company said V4.1Flash1017 is available through its website, mobile app and API. DeepSeek described the release as a minor update that improves performance and cost. The company’s reported benchmark scores rose on two tests: V4.1Flash1017 scored 77.0 on DeepSWE v1.1, compared with 74.2 for the earlier model, and 60.5 on AutomationBench V1.0.6, up from 54.8. Those figures were reported by DeepSeek, and the material available does not establish independent testing of the updated model’s performance across broader workloads. The scores indicate gains on the named benchmarks, but do not by themselves demonstrate how the system will perform in all real-world uses. Alongside the model update, DeepSeek opened its Search and Fetch APIs to all users. The company set the price at 0.75 per 1,000 calls for each endpoint. DeepSeek said the APIs run on its search engine and can be used to add search capabilities to applications. The company also said the updated model generates shorter responses,...