Tencent model product
Hy3
8B MTP layer, developed by the Tencent Hy Team.
Updated Aug 17, 2026. Default version: Hy3
LLMBoard Score70.7Hy3
Coverage80%25 benchmark families
Context window256KTokens
Official input priceN/AOfficial price unavailable
Hy3 Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Hy3 Benchmark Results
Benchmark scores for Hy3.
Sort by
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkCL-bench | Score23.8% | Rank01 | Participants2 | Percentile100.0% | EvidenceC | Evaluated |
| BenchmarkCL-bench (Life) | Score17.0% | Rank01 | Participants1 | Percentile100.0% | EvidenceC | Evaluated |
| BenchmarkCMT-Benchmark | Score37.9% | Rank01 | Participants1 | Percentile100.0% | EvidenceC | Evaluated |
| BenchmarkHorizonMath | Score7.1% | Rank01 | Participants3 | Percentile100.0% | EvidenceC | Evaluated |
| BenchmarkHumanity's Last Exam (no tools, text-only) | Score47.0% | Rank01 | Participants2 | Percentile100.0% | EvidenceC | Evaluated |
| BenchmarkPHYBench | Score77.4% | Rank01 | Participants1 | Percentile100.0% | EvidenceC | Evaluated |
| BenchmarkAA-LCR | Score73.4% | Rank02 | Participants18 | Percentile94.1% | EvidenceC | Evaluated |
| BenchmarkArXivMath | Score52.2% | Rank02 | Participants2 | Percentile0.0% | EvidenceC | Evaluated |
| BenchmarkFrontierScience Olympiad | Score74.8% | Rank03 | Participants3 | Percentile0.0% | EvidenceC | Evaluated |
| BenchmarkHumanity's Last Exam (with tools, text-only) | Score53.2% | Rank03 | Participants3 | Percentile0.0% | EvidenceC | Evaluated |
| BenchmarkIMO-AnswerBench | Score90.0% | Rank03 | Participants20 | Percentile89.5% | EvidenceC | Evaluated |
| BenchmarkSkillsBench | Score55.3% | Rank03 | Participants8 | Percentile71.4% | EvidenceC | Evaluated |
| BenchmarkSuperChem | Score54.9% | Rank03 | Participants3 | Percentile0.0% | EvidenceC | Evaluated |
| BenchmarkUSAMO 2026 | Score30.24 points | Rank03 | Participants3 | Percentile0.0% | EvidenceC | Evaluated |
| BenchmarkWildClawBench | Score53.6% | Rank03 | Participants5 | Percentile50.0% | EvidenceC | Evaluated |
| BenchmarkClaw-Eval | Score68.5% | Rank04 | Participants14 | Percentile76.9% | EvidenceC | Evaluated |
| BenchmarkDeepSearchQA | Score91.0% | Rank04 | Participants9 | Percentile62.5% | EvidenceC | Evaluated |
| BenchmarkFrontierScience Research | Score21.3% | Rank04 | Participants4 | Percentile0.0% | EvidenceC | Evaluated |
| BenchmarkWideSearch | Score76.4% | Rank04 | Participants10 | Percentile66.7% | EvidenceC | Evaluated |
| BenchmarkMathArena Apex | Score38.7% | Rank05 | Participants8 | Percentile42.9% | EvidenceC | Evaluated |
| BenchmarkAPEX-Agents | Score25.6% | Rank08 | Participants8 | Percentile0.0% | EvidenceC | Evaluated |
| BenchmarkMCP Atlas | Score79.1% | Rank08 | Participants33 | Percentile78.1% | EvidenceC | Evaluated |
| BenchmarkNL2Repo | Score45.6% | Rank08 | Participants17 | Percentile56.3% | EvidenceC | Evaluated |
| BenchmarkDeepSWE | Score28.0% | Rank10 | Participants11 | Percentile10.0% | EvidenceC | Evaluated |
| BenchmarkSWE-bench Multilingual | Score75.8% | Rank10 | Participants38 | Percentile75.7% | EvidenceC | Evaluated |
| BenchmarkBrowseComp | Score84.2% | Rank14 | Participants62 | Percentile78.7% | EvidenceC | Evaluated |
| BenchmarkTerminal-Bench 2.1 | Score71.7% | Rank17 | Participants28 | Percentile40.7% | EvidenceC | Evaluated |
| BenchmarkGPQA | Score90.4% | Rank21 | Participants239 | Percentile91.6% | EvidenceC | Evaluated |
| BenchmarkSWE-Bench Pro | Score57.9% | Rank21 | Participants50 | Percentile59.2% | EvidenceC | Evaluated |
| BenchmarkSWE-Bench Verified | Score78.0% | Rank22 | Participants111 | Percentile80.9% | EvidenceC | Evaluated |
| BenchmarkToolathlon | Score48.5% | Rank22 | Participants37 | Percentile41.7% | EvidenceC | Evaluated |
Hy3 Arena Results
Preference and agent-evaluation results for the default version.
Sort by
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenawebdev | Categoryoverall | Rank24 | Rating / score1521.7 | Votes2,284 | ObservationsN/A | Result date |
| Arenaagent praise complaint | Categoryoverall | Rank26 | Rating / score0.0 | VotesN/A | Observations2.6K | Result date |
| Arenaagent bash recovery steps | Categoryoverall | Rank29 | Rating / score0.0 | VotesN/A | Observations16K | Result date |
| Arenaagent | Categoryoverall | Rank32 | Rating / score-0.0 | VotesN/A | Observations620.8K | Result date |
| Arenaagent task outcome explicit | Categoryoverall | Rank36 | Rating / score-0.0 | VotesN/A | Observations8.3K | Result date |
| Arenaagent steerability | Categoryoverall | Rank42 | Rating / score-0.1 | VotesN/A | Observations10.3K | Result date |
| Arenaagent tool hallucination | Categoryoverall | Rank45 | Rating / score-0.0 | VotesN/A | Observations583.7K | Result date |
| Arenatext style control | Categoryoverall | Rank53 | Rating / score1456.9 | Votes4,650 | ObservationsN/A | Result date |
| Arenatext | Categoryoverall | Rank57 | Rating / score1441.5 | Votes4,650 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryoverall | Rank57 | Rating / score1448.2 | Votes4,650 | ObservationsN/A | Result date |
| Arenawebdev | Categoryoverall | Rank83 | Rating / score1356.1 | Votes1,394 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryoverall | Rank117 | Rating / score1415.8 | Votes6,621 | ObservationsN/A | Result date |
| Arenatext style control | Categoryoverall | Rank123 | Rating / score1412.8 | Votes6,624 | ObservationsN/A | Result date |
| Arenatext | Categoryoverall | Rank127 | Rating / score1405.1 | Votes6,624 | ObservationsN/A | Result date |
Hy3 Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- N/A
- Official provider
- N/A
- Lowest third-party
- From $0.066 input, $0.26 output per 1M via NanoGPT
- Tracked offerings
- 12
Sort by
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderTencent TokenHub | Provider model IDhy3 | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context256K | Updated |
| ProviderTencent Token Plan | Provider model IDhy3 | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context256K | Updated |
| ProviderNanoGPT | Provider model IDtencent/hy3 | Regionglobal | Input / 1M$0.066 | Output / 1M$0.26 | Context262.1K | Updated |
| ProviderOpenRouter | Provider model IDtencent/hy3 | Regionglobal | Input / 1M$0.132 | Output / 1M$0.528 | Context262.1K | Updated |
| ProviderDeep Infra | Provider model IDtencent/Hy3 | Regionglobal | Input / 1M$0.14 | Output / 1M$0.58 | Context262.1K | Updated |
| ProviderHugging Face | Provider model IDtencent/Hy3 | Regionglobal | Input / 1M$0.14 | Output / 1M$0.58 | Context262.1K | Updated |
| ProviderOpenCode Go | Provider model IDhy3 | Regionglobal | Input / 1M$0.14 | Output / 1M$0.58 | Context256K | Updated |
| ProviderVercel AI Gateway | Provider model IDtencent/hy3 | Regionglobal | Input / 1M$0.14 | Output / 1M$0.58 | Context262.1K | Updated |
| ProviderEden AI | Provider model IDdeepinfra/tencent/Hy3 | Regionglobal | Input / 1M$0.14 | Output / 1M$0.58 | Context262.1K | Updated |
| ProviderLLM Gateway | Provider model IDhy3 | Regionglobal | Input / 1M$0.14 | Output / 1M$0.58 | Context262.1K | Updated |
| ProviderKilo Gateway | Provider model IDtencent/hy3 | Regionglobal | Input / 1M$0.14 | Output / 1M$0.58 | Context262.1K | Updated |
| ProviderCrossModel | Provider model IDtencent/hy3 | Regionglobal | Input / 1M$0.16 | Output / 1M$0.64 | Context262.1K | Updated |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Hy3 Runtime Performance
Provider-specific output speed and catalog latency for Hy3. Runtime does not affect the capability score.
No runtime data
Browse runtime rankingsNo provider-specific speed or latency record is linked to the default version yet.
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Hy3 Specifications
Technical details for the model's default version.
- Version
- Hy3
- Released
- Jul 6, 2026
- Knowledge cutoff
- Unknown
- Parameters
- 295B
- Context window
- 256K
- Max output
- 64K
- Inputs
- text
- Outputs
- text
- Open weights
- Yes
- License
- Apache 2.0
Hy3 Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
Sort by
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionHy3 | Released | LLMBoard70.7 | Parameters295B | Context256K | Max output64K | Open weightsYes | LicenseApache 2.0 |
Hy3 vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to Hy3
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
MI
MiniMax M3
MiniMax
OP
GPT-5.2
OpenAI
PO
Laguna S 2.1
Poolside
AN
Claude Opus 4.5
Anthropic
ZA
GLM 5.1
Zhipu AI
OP
GPT-5.2-Pro
OpenAI
What is Hy3?
Key information about Hy3 and its available data.
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 data and RL, gathering feedback from 50+ products.
Hy3 outperforms similar-size models and rivals flagship open-source models with 2-5x the parameters, with strong gains in reasoning, agentic, and long-context tasks.
It uses 80 layers (plus 1 MTP layer), 64 GQA attention heads (8 KV heads, head dim 128), a 4096 hidden size, 192 experts with top-8 activated, a 256K context window, and BF16 precision.
Hy3 is a hybrid-thinking model supporting configurable reasoning effort (no_think, low, high), and emphasizes production-grade tool-call and output-format stability, reduced hallucination, and reliable multi-turn intent tracking.
Data as of 2026-08-17.
FAQ
Common questions about Hy3.
When was Hy3 released?
Hy3's default version was released on Jul 6, 2026.
How much does Hy3 cost?
No official standard PAYG price is currently available for Hy3. The lowest tracked third-party offer starts at $0.066 input and $0.26 output via NanoGPT.
Who created Hy3?
Hy3 was created by Tencent.
What is the context window for Hy3?
The default version has a 256K token context window.
Is Hy3 open weight?
Yes. The default version is marked as open weight under Apache 2.0.
How many API providers offer Hy3?
12 provider offerings are linked to the default version.
What models should I compare Hy3 with?
Nearby ranked alternatives include Sakana Namazu, GPT-5.2-Pro, MiniMax M3.