Tencent · Hunyuan
Think it through. Put it to work.
A new reasoning model for your Offloop agent. Bring Hunyuan 4 to the questions, code and projects that need more than a quick answer.
Available in Offloop preview
Official launch film · Tencent AI · November 4, 2026 · 1 min 10 sec
The film shows Tencent’s newly released model. Features and model versions available in Offloop may differ; check your workspace for current access.
Context reflects the Hunyuan 4 configuration in Offloop. Availability and limits may change during preview.
Explore 12 benchmarks from Tencent’s release report. Switch the task, compare every reported model, and inspect the numbers behind the chart.
Tencent-reported · November 4, 2026
Choose a benchmark
Higher is better · 0–100 scale
Hover or tap a bar for its score. The complete data is also available below.
| Model | Reported score |
|---|---|
| Hy4 | 44.4 |
| GLM 5.3 | 42.9 |
| DeepSeek V4.1 Flash 1017 | 40.0 |
| Kimi K3.1 Pro | 66.8 |
| GPT 6.1 Sol | 55.5 |
| Claude Opus 5.5 | 66.4 |
Values reproduce Tencent’s launch summary chart. Tencent reports the highest available reasoning setting for each model; harnesses and evaluation conditions vary by benchmark. These are provider-reported results, not Offloop measurements. The full appendix may also show previously published scores alongside Tencent’s own tests.
This Agent Arena snapshot places Hy4 on the plotted Pareto frontier: a view of score alongside the cost of completing a task.
Snapshot supplied for this page. Cost uses the chart’s trailing 14-day window. The capture date and full methodology are not included in the image; this is not a live leaderboard or Offloop pricing.
Give your agent a real task, useful context and a clear outcome. Here are three places to begin.
01
Bring a tricky implementation question or a change that needs careful thought. Ask your agent to explain its approach and the tradeoffs before moving ahead.
A prompt to try
“Review this change. Explain the edge cases and suggest the smallest fix.”
02
Turn a broad question into a focused research task. Ask for sources, a comparison and a recommendation you can evaluate.
A prompt to try
“Research these three options. Compare the tradeoffs and link the sources behind your recommendation.”
03
Give scattered notes a direction. Work with your agent to turn project context into a brief, a plan or a draft ready for your review.
A prompt to try
“Turn these project notes into a decision brief: what we know, what is missing and what to do next.”
Suggested starting prompts, not recorded model outputs. Available tools depend on your agent’s setup and permissions.
Choose the agent you want to work with.
Choose Hunyuan 4 in your agent’s model settings, where available.
Share the task, the context and what a good result should look like.
Start with a question. Take it somewhere useful.
Open Offloop
