Tencent Hunyuan released its next-generation large language model Hy4 on Nov 4, saying the model has entered the top tier of open-source models for the first time. According to kaopunews, that is the core information available so far on Hy4's positioning.

A day earlier, Google released its next-generation AI model Gemini 4 Carbon. According to Tom's Hardware, the model topped intelligence rankings, narrowly leading Claude Opus 5.5, and was described as the frontier model "least likely to hallucinate." The benchmark results were provided by the vendor.

Two Models Debut Back to Back

The release information for Tencent Hunyuan's Hy4 was brief. Public materials mention only a "next-generation large language model" and the positioning of "entering the top tier of open-source models for the first time," without disclosing specifics such as parameter count, benchmark scores or open-source license terms.

Coverage of Google Gemini 4 Carbon offered more comparative detail. Tom's Hardware said it "narrowly led" Claude Opus 5.5 in intelligence rankings and highlighted its performance in reducing hallucinations. However, the benchmark data all came from Google and has not been independently verified by third parties.

Two Paths: Open Source and Closed Source

Based on public positioning, Hy4 emphasizes the "open-source model tier," while Gemini 4 Carbon continues Google's frontier closed-source model approach. The near-simultaneous releases offer a reference point for observing the relative standing of domestic open-source models and global frontier closed-source models.

The claim that Hy4 has "entered the top tier of open-source models for the first time" is currently only Tencent Hunyuan's own statement, with no independent evaluator or third-party ranking data to corroborate it.

Gemini 4 Carbon's lead was described as "narrow," suggesting the gap between frontier models may be shrinking.

What to Watch Next

  • Whether Tencent Hunyuan discloses Hy4's benchmark scores, parameter count and open-source license terms;
  • Independent verification results for Hy4 and Gemini 4 Carbon from third-party evaluators;
  • Hy4's actual adoption in the open-source community and how it compares with similar open-source models;
  • Whether Gemini 4 Carbon's anti-hallucination performance can be reproduced in independent tests.

This information will determine the two models' actual competitiveness and the real value of the "top tier of open-source models" label.