Tencent Hunyuan Hy3 Preview Launches, Token Usage Jumps 10x in 2 Weeks, Tops OpenRouter

Two weeks after its launch, Tencent Hunyuan's Hy3preview has seen rapid adoption across the developer ecosystem, with total token calls exceeding ten times those of its predecessor, Hy2. Monitoring data reveals that code writing and intelligent agent development have become primary growth drivers, boosting combined call volume for internal tools like WorkBuddy, Codebuddy, and Qclaw by over 16.5 times.
Hy3preview also leads in the external developer market. According to OpenRouter's weekly data, the model ranked first in both total token call volume and market share, while also taking the top spot for call popularity in advanced use cases such as programming assistance and tool invocation.
The head of Tencent Hunyuan noted that the initial goal behind Hy3preview was to maximize practicality. By offering a free trial on OpenRouter, the team gathered substantial developer feedback in real-world, complex settings. This community-driven, open platform testing approach is becoming a key driver for rapid iteration of large language models.
Industry analysts view Hy3preview's advances in code and tool call scenarios as a shift for domestic large models—from simple conversational abilities into "action-oriented" tools with real productivity value. As call volumes grow exponentially, massive real-world data feedback will further refine the model's logical reasoning and task execution limits, accelerating the penetration of large models into enterprise applications and automated workflows across China.
Related article
Superintelligence Arrives: Do We Let It In?
Loading the player…AI companies have been talking about superintelligent AI like it’s inevitable, but recent safety incidents like OpenAI’s Hugging Face breach are demonstrating the potential dangers of deploying AI systems that are more capable than
iFLYTEK Unveils Spark X2.5 General Large Model
On September 1, iFlytek will open-source two edge-focused large language models, Xinghuo X2.5-4B and Xinghuo X2.5-1.7B, according to the latest official announcement. Both models natively support a context window of up to 1 million tokens. They deliv
Meta Removes AI Photo Editing Feature Following User Backlash
Meta, the social media giant, is once again embroiled in a public debate regarding the delicate balance between artificial intelligence and user privacy. According to TechCrunch, Meta’s Superintelligence Labs introduced a new AI image generator, Muse
Related Special Topic Recommendations
Comments (0)
0/500

Two weeks after its launch, Tencent Hunyuan's Hy3preview has seen rapid adoption across the developer ecosystem, with total token calls exceeding ten times those of its predecessor, Hy2. Monitoring data reveals that code writing and intelligent agent development have become primary growth drivers, boosting combined call volume for internal tools like WorkBuddy, Codebuddy, and Qclaw by over 16.5 times.
Hy3preview also leads in the external developer market. According to OpenRouter's weekly data, the model ranked first in both total token call volume and market share, while also taking the top spot for call popularity in advanced use cases such as programming assistance and tool invocation.
The head of Tencent Hunyuan noted that the initial goal behind Hy3preview was to maximize practicality. By offering a free trial on OpenRouter, the team gathered substantial developer feedback in real-world, complex settings. This community-driven, open platform testing approach is becoming a key driver for rapid iteration of large language models.
Industry analysts view Hy3preview's advances in code and tool call scenarios as a shift for domestic large models—from simple conversational abilities into "action-oriented" tools with real productivity value. As call volumes grow exponentially, massive real-world data feedback will further refine the model's logical reasoning and task execution limits, accelerating the penetration of large models into enterprise applications and automated workflows across China.
Superintelligence Arrives: Do We Let It In?
Loading the player…AI companies have been talking about superintelligent AI like it’s inevitable, but recent safety incidents like OpenAI’s Hugging Face breach are demonstrating the potential dangers of deploying AI systems that are more capable than
iFLYTEK Unveils Spark X2.5 General Large Model
On September 1, iFlytek will open-source two edge-focused large language models, Xinghuo X2.5-4B and Xinghuo X2.5-1.7B, according to the latest official announcement. Both models natively support a context window of up to 1 million tokens. They deliv
Meta Removes AI Photo Editing Feature Following User Backlash
Meta, the social media giant, is once again embroiled in a public debate regarding the delicate balance between artificial intelligence and user privacy. According to TechCrunch, Meta’s Superintelligence Labs introduced a new AI image generator, Muse





Home






