Tencent Cloud Sets Mid-July Launch for DeepSeek-V4 Official Release
Tencent Cloud has officially announced that the complete version of DeepSeek-V4 is scheduled for release in mid-July. This "direct-from-source" model will become available simultaneously on the company's large model service platform TokenHub and its agent development platform.
Comprehensive Functional and Performance Enhancements
The upcoming official release represents a substantial upgrade rather than a minor iteration. The Tencent Cloud engineering team has conducted deep optimizations targeting improved performance across multiple dimensions including complex task handling, multi-turn conversations, and code generation.
Key improvements over previous versions include:
- Enhanced Inference Efficiency: Reduced response latency and improved stability under high concurrency
- Better Context Understanding: Refined capabilities for processing long texts and following complex instructions
- Higher Output Quality: More impressive performance in creative writing and technical documentation generation
Updated Service and Pricing Framework
Alongside the version launch, Tencent Cloud will implement adjustments to its service framework. Most notably, the company will introduce a peak-valley pricing mechanism. This dynamic pricing model is designed to better align with actual usage patterns, offering more competitive rates during periods of lower demand.
According to Tencent Cloud, the pricing revision aims to optimize computational resource allocation and ensure consistent service quality during peak usage times. For developers and enterprise users, this change allows for cost optimization by scheduling resource-intensive tasks during more economical time windows.
The platform advises existing users to monitor official channels for detailed pricing adjustments and migration guidelines. The new billing system is expected to take effect concurrently with the official model release.