Tencent Hy4 preview
Tencent's open-weight 770B sparse MoE with 49B active parameters, a 1M context window, Apache 2.0 weights, and publisher-reported coding and reasoning results Tencent places at the open-source frontier.
Routable on MiniRouter
Use tencent/hy4-preview via the API →- Architecture
- 770B MoE · 49B active
- 256 routed plus one shared expert per layer; gated sparse attention
- Context
- 1M
- 960K maximum input, 64K maximum output
- Weights
- Released
- Apache 2.0 on Hugging Face, ModelScope, GitCode, and CNB
- Reasoning
- High by default
- A no_think effort setting disables chain-of-thought
Publisher pricing
Pricing below comes from the release announcement and does not represent a MiniRouter route or price.
| Phase | Input / 1M | Output / 1M | Blended / 1M | Effective | Source |
|---|---|---|---|---|---|
| Tencent Cloud API | $0.83 | $2.50 | $1.25 | Aug 28, 2026 – open-ended | Tencent Cloud ↗ |
Publisher-reported benchmarks
Results are grouped by unit. Percentage, Elo, and task-count results never share a scale.
| Benchmark | Reported result | Reported by | As of | Source |
|---|---|---|---|---|
| GPQA Diamond | 92.3 | Tencent | Aug 28, 2026 | Announcement ↗ |
| Terminal-Bench 2.1 | 85.4 | Tencent | Aug 28, 2026 | Announcement ↗ |
| SWE-Bench Pro | 65.7 | Tencent | Aug 28, 2026 | Announcement ↗ |
| SWE-Bench Multilingual | 82.9 | Tencent | Aug 28, 2026 | Announcement ↗ |
| Deep SWE | 64.3 | Tencent | Aug 28, 2026 | Announcement ↗ |
| Apex Agents | 37.1 | Tencent | Aug 28, 2026 | Announcement ↗ |
| SkillsBench v1.1 | 62.9 | Tencent | Aug 28, 2026 | Announcement ↗ |
| Toolathlon Verified | 74.1 | Tencent | Aug 28, 2026 | Announcement ↗ |
- — Tencent calls this a preview and says pre-training and post-training still have room to improve.
- — Tencent's blind expert evaluation against GLM 5.3 and Kimi K3 is an internal study and is not listed as a benchmark here.
- — Tencent reports the benchmark figures; MiniRouter has not reproduced them.
Independent measurements
Artificial Analysis runs its own evaluations and timing against each model's first-party API. These are the only numbers on this page not reported by the publisher.
Not yet evaluated
Artificial Analysis has not published measurements for Tencent Hy4 preview as of Sep 2, 2026. Publisher claims above stand alone until they do.
Source: Artificial Analysis ↗, read Sep 2, 2026. Measured on the model's first-party API, not a MiniRouter route.
What is not independently confirmed
- Every figure on this page is self-reported by the cited publisher. MiniRouter did not run these benchmarks.
- Benchmark harnesses are not standardised. Results sharing a name across vendors have not been confirmed to use identical runs.
- Publisher pricing can change, and introductory phases have stated end dates.
- Tencent Hy4 preview is routable through MiniRouter today; live pricing on its model page is authoritative over the publisher figures here.
Related coverage
Other release dossiers and the current MiniRouter catalog.
Sources
- Read at Tencent ↗
Tencent Hy4 preview model card
Published Aug 28, 2026
Editorial reference revised 2026-09-02. Catalog status comes from the generated routing snapshot.