Hy4 Preview
Public AI IQ model profile with source-backed benchmark results and derived ranking context.
Hy4 Preview API pricing — full price table, workload cost examples, and cheaper alternatives.
Release and Capabilities
| Public release | 2026-08-28 |
|---|---|
| Modalities | text |
| Context window | 1,048,576 tokens |
Open weights on Hugging Face and hosted access through Tencent products, Tencent Cloud TokenHub, and OpenRouter
Publisher-reported capabilities
- Hy4 Preview is Tencent's next-generation flagship mixture-of-experts model with 770B total parameters and 49B active parameters.
- It supports more than 1M tokens of context and targets coding, office work, analysis, game development, and scientific research.
Limitations and availability caveats
- Tencent describes this as an early preview with headroom remaining in pre-training and post-training.
- Known issues include reasoning longer than necessary on complex tasks and a tendency to over-verify its work.
- Provider-reported launch evaluations are not substituted for independent leaderboard results in sparse benchmark files.
Official release and model references
Measured Coverage and Estimates
Hy4 has 4 of 39 tracked IQ benchmark inputs across 3 of 7 dimensions. Raw benchmark tables contain source-backed results only. Derived IQ combines all six dimensions using the shared scoring methodology, including estimates where direct coverage is incomplete.
Among dimensions with direct benchmark coverage, this profile's highest estimated scores are Programmatic Reasoning (IQ 138) and Mathematical Reasoning (IQ 136). These dimension estimates can include statistical adjustments or predecessor priors; they are not raw benchmark percentages or claims of real-world task success.
Benchmark versions, evaluation harnesses, and coverage affect comparisons; a missing result is not a score of zero. See the scoring methodology for estimation and uncertainty limits.
IQ Dimensions
| Dimension | IQ | Coverage |
|---|---|---|
| Abstract Reasoning | 89 | 0/5 |
| Mathematical Reasoning | 136 | 1/5 |
| Academic Reasoning | 100 | 0/6 |
| Programmatic Reasoning | 138 | 2/6 |
| Computer Use | 125 | 1/5 |
| Reliability | 101 | 0/6 |
| Knowledge Work | 100 | 0/6 |
Benchmark Results
| Benchmark | Score | Dimension |
|---|---|---|
| Arena.ai Agent Arena | 2.46 | Computer Use |
| IOI V2 | 59.333 | Programmatic Reasoning |
| ProgramBench | 14.5 | Programmatic Reasoning |
| ProofBench 1.1 | 75 | Mathematical Reasoning |
Compare
Frequently Asked Questions
- What is the IQ of Hy4 Preview?
- Hy4 Preview has an estimated IQ of 113, which ranks #60 out of 164 public models on the AI IQ leaderboard. The score combines seven equally weighted reasoning dimensions using measured benchmark results and conservative estimates for missing coverage.
- How much does Hy4 Preview cost?
- Hy4 Preview costs $0.834 per million input tokens and $2.50 per million output tokens. Adjusted for how many tokens it actually consumes on real tasks, its effective cost works out to $3.19 per million input-plus-output tokens.
- Is Hy4 Preview open source?
- Yes — Hy4 Preview is an open-weights model: its trained weights are published, so it can be downloaded and self-hosted.
- When was Hy4 Preview released?
- Hy4 Preview was publicly released on August 28, 2026.
- How large is the context window of Hy4 Preview?
- Hy4 Preview supports a context window of 1,048,576 tokens — the amount of text it can consider in a single request.