TL;DR
- The mysterious Hy3 LLM is topping OpenRouter Model Rankings by a large margin.
- Hy3 preview is available from the OpenRouter API at a stated price of $0.066/1M tokens input.
- Hy3's model quality is on par with other Chinese models, but not close to models like Claude Opus 4.7 and GPT 5.5.
- Hy3's popularity is not limited to agentic coding domains.
The OpenRouter Model Rankings have recently been topped by the mysterious Hy3 LLM, with two new models beating LLM darling Claude in terms of token usage by more than 50%. According to the OpenRouter rankings retrieved on May 25, 2026, Hy3 preview and DeepSeek Flash V4 are the top two models. Hy3's popularity is puzzling, given its relatively unknown status and benchmark results that are not favorable compared to other Chinese open-source models.
What the data shows
The OpenRouter rankings provide insight into how users interact with LLMs, with Hy3 preview and DeepSeek Flash V4 leading the pack. The data shows that Hy3 preview is available at a stated price of $0.066/1M tokens input, which is cheaper than DeepSeek V4 Flash at $0.10/1M tokens input. This price difference may be a factor in Hy3's popularity, especially given the rising cost of LLMs and coding agents.
- Hy3 preview is topping the OpenRouter Model Rankings.
- DeepSeek Flash V4 is the second most popular model.
- The price of Hy3 preview is $0.066/1M tokens input.
What this means for ai readers
The popularity of Hy3 preview has implications for AI readers, particularly those interested in LLMs and their applications. The fact that Hy3's model quality is on par with other Chinese models, but not close to models like Claude Opus 4.7 and GPT 5.5, suggests that there may be other factors at play in its popularity. Additionally, Hy3's availability at a lower price point than other models may make it an attractive option for users looking for a cost-effective solution.
Hy3's popularity is not limited to agentic coding domains, which may indicate that it has a broader range of applications.
What to do right now
For those interested in exploring Hy3 preview, the model is available through the OpenRouter API. Users can test the model and see how it performs in their specific use cases. Additionally, the OpenRouter rankings provide a useful resource for comparing the performance of different LLMs and identifying trends in the field.
- Test Hy3 preview through the OpenRouter API.
- Compare the performance of different LLMs using the OpenRouter rankings.
Bottom line
The mysterious Hy3 LLM is topping the OpenRouter Model Rankings, and its popularity is likely due to a combination of factors, including its price point and performance. While Hy3's model quality may not be on par with other leading models, its availability at a lower price point and its broader range of applications make it an attractive option for users.
Frequently asked questions
Q: What is the price of Hy3 preview?
Hy3 preview is available at a stated price of $0.066/1M tokens input.
Q: How does Hy3's model quality compare to other models?
Hy3's model quality is on par with other Chinese models, but not close to models like Claude Opus 4.7 and GPT 5.5.
Q: What is the source of the data on Hy3's popularity?
The data on Hy3's popularity comes from the OpenRouter Model Rankings, which were retrieved on May 25, 2026.
Q: Is Hy3's popularity limited to agentic coding domains?
No, Hy3's popularity is not limited to agentic coding domains, and it may have a broader range of applications.
Sources
Related reading
- Real-time LLM Inference on Standard GPUs: 3k tokens/s per request
- Anthropic Launches Claude Fable 5 and Mythos 5: A New Model Class Above Opus
- If you're an LLM
Originally published on AI at Crescevo — subscribe free for more.
Top comments (0)