Hy3: API Pricing, Context Window & Benchmarks
Hy3 is a language model from Tencent, released in July 2026.
Hy3 is a 295B-parameter Mixture-of-Experts (MoE) model with 21B active parameters and a 3.8B MTP layer, developed by the Tencent Hy Team. Following the Hy3 Preview launch in late April, the team scaled up post-training with higher-quality
Hy3 benchmarks
Capability tiers
Standing within each category, adjusted for leaderboard depth.
Real tasks performance
High-confidence performance for Hy3 across real-world prompt categories. Only 95% intervals at most 4 points wide are shown.
Performance by conversation depth
How Hy3 holds up as conversations get longer.
Quality Tracker
Hy3 Performance Across Datasets
Scores sourced from the model's scorecard, paper, or official blog posts
llm-stats.com - Sun Aug 30 2026Hy3 model size
Hy3 has 295 billion parameters. See how it compares to other models in the same parameter range.
Hy3 latency
Hy3 time to first token, sustained output throughput, and failed-request rate from live API traffic over the trailing 7 days.
Hy3 examples
Recent arena outputs from Hy3, picked from the highest-ranked matchups.
Hy3 license
Hy3 is released under the Apache 2.0 license, which permits commercial use, has 295.0B parameters.
- License
- Apache 2.0
- Commercial use allowed
- Parameters
- 295.0B
Apache License 2.0 - allows commercial use
Hy3 resources
Official sources for Hy3: official launch post, source repository, model weights.
Hy3 vs other models
The most-compared alternatives to Hy3 are GLM-5.1, Claude Opus 4.6, Gemini 3 Flash. Open any pair side-by-side for benchmarks, pricing, context, and latency.
Models like Hy3
Models ranked just above and below Hy3 by LLM Stats score.
FAQ
Common questions about Hy3.