Hy3
Tencent · released Jul 6, 2026
Hy3 is a 295B-parameter Mixture-of-Experts model from Tencent (21B active, 192 experts with top-8 routing) built for reasoning, agentic workflows, and real-world production use. It supports a configurable reasoning effort:...
Specification
- Context window
- 262K
- Max output
- 128K
- Knowledge cutoff
- Not stated
- Parameters
- 298.8B
- Licence
- apache-2.0
- Serving providers
- 5
- Moderated
- No
- Uptime
- 100.0%
Intelligence
25.3
77th percentile
Coding
58.8
Coding Index
Agentic
24.1
Agentic Index
Output speed
66 t/s
Median across providers
Latency
1.66s
Time to first token
Cost per task
$0.05
Estimated
Benchmarks
Where the score comes from
The Intelligence Index is a composite. These are the underlying evaluations this model was actually measured on.
Evaluation scores
Percentage correct · higher is better
- GPQA Diamond89.7%
- AA-LCR (long context)79.0%
- SciCode48.6%
- Humanity's Last Exam33.5%
- τ²-bench (Telecom)22.9%
An evaluation missing from this list was not run for this model — it is not a zero.
View as table
| Evaluation | Score |
|---|---|
| GPQA Diamond | 89.7% |
| AA-LCR (long context) | 79.0% |
| SciCode | 48.6% |
| Humanity's Last Exam | 33.5% |
| τ²-bench (Telecom) | 22.9% |
Against its peers
Intelligence Index · this model highlighted, nearest peers in grey
- Inkling SmallThinking Machines25.7
- Inkling Small (free)Thinking Machines25.7
- Grok 4.20xAI25.7
- Hy3Tencent25.3
- Qwen3.7 PlusQwen25.2
- InklingThinking Machines25.0
- Inkling (free)Thinking Machines25.0
- Grok 4.3xAI24.9
Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.
View as table
| Model | Intelligence |
|---|---|
| Inkling Small | 25.7 |
| Inkling Small (free) | 25.7 |
| Grok 4.20 | 25.7 |
| Hy3 | 25.3 |
| Qwen3.7 Plus | 25.2 |
| Inkling | 25.0 |
| Inkling (free) | 25.0 |
| Grok 4.3 | 24.9 |
Percentile among all indexed models
Pricing
What it costs to run
List prices per million tokens, plus what one representative task works out to.
List price
- Input / 1M tokens
- $0.132
- Output / 1M tokens
- $0.528
- Cached input / 1M
- $0.033
- Blended 3:1
- $0.231
One task, estimated
$0.05
- Input tokens
- 50,000
- Output tokens
- 80,000
- Profile
- Reasoning
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).
Serving providers
Speed and latency figures are medians across these providers, so a widely-served model reports a blend rather than any single endpoint.
Arena
Head-to-head generation quality
Elo from pairwise judgements, broken out by the kind of thing the model was asked to build.
- Overall Elo
- 1,183
- Win rate
- 41.3%
- Strongest at
- Websites
- Tournaments
- 2,409
Elo by category
Dot position on a 1,125–1,200 scale · Elo has no meaningful zero
- 1,183
- 1,176
- 1,165
- 1,139
- 1,125
Agent categories (full-stack apps, mobile apps) are only scored for models tested in agent harnesses.
View as table
| Category | Elo | Win rate |
|---|---|---|
| Websites | 1183 | 41.3% |
| 3D scenes | 1176 | 44.0% |
| UI components | 1165 | 40.2% |
| Game development | 1139 | 38.6% |
| Data visualisation | 1125 | 36.1% |