DeepSeek has moved the updated 0813 build of DeepSeek-V4-Pro into general availability across its app, website and API. The rollout began on 12 August 2026, and the company recorded the GA release in its changelog on 13 August. This is not a new V5 or R2 model; it is the updated production build of V4 Pro. API users do not need to change their requests to a new public model name. It remains available as deepseek-v4-pro, while the app and web service receive the same update. DeepSeek says the release focuses on agent workloads: planning multi-step tasks, using tools, inspecting code and completing workflows rather than producing only a single response. Company-published benchmark results report gains in agentic coding, terminal use, automation and tool-assisted reasoning. These are vendor results, so independent testing and real user experience remain important measures of practical quality. Early reactions have been mixed: some users praised its coding ability, while others said the improvement over V4 Flash did not always justify the extra cost. V4 Pro retains the foundation of the V4 family introduced in April, including a context window of up to one million tokens and a focus on reasoning and coding. The 0813 update also adds native Responses API support, which can simplify agents that combine search, code and other tools. DeepSeek also announced that revised API pricing will take effect at 16:00 UTC on 16 August, using peak and off-peak periods. Developers therefore need to evaluate more than benchmark scores: quality, speed and the cost of their own workload all matter. Somali developers building chatbots, coding assistants or automation can use the release through the deepseek-v4-pro model name, but should review the new pricing before running high-volume services.