OpenClaw adopted DeepSeek’s new V4 Flash model as the default for its AI agent and added DeepSeek’s flagship V4 Pro models to its catalogue on Sunday, after DeepSeek said its V4 line was tuned for multi-step tasks. [1]
DeepSeek announced on Friday that its V4 models were optimized to stay consistent during longer agent workflows and were built for mainstream tools including Anthropic’s Claude Code, OpenClaw and Tencent Holdings’ CodeBuddy. [1]
OpenClaw also integrated extra capabilities into its agent, including Google Meet, as it expanded support around the DeepSeek models. [1]
Huawei Technologies said it now fully supports DeepSeek V4 inference on its Ascend chips and supernode systems, adding a hardware layer to the rollout that has drawn wide attention in the tech community. [1]
The latest change follows DeepSeek’s May 1 announcement of its V4 models and OpenClaw’s May 3 update to make V4 Flash the default and bring in V4 Pro support. [1]
