GPQA DiamondVerified
89.7%
GPQA Diamond (AA run)
Models / Other
Hy3
Other · Released 2026-07-06
Tencent Hunyuan Hy3. Superseded as the current Hunyuan preview by Hy4 preview (28 Aug 2026, 770B / 49B, $0.834 / $2.501).
Frontier model tracked on Artificial Analysis. Use the benchmark ledger below for sourced scores; we do not infer numbers beyond published rows.
Composite62
Context128K
Input / 1MFree
Output / 1MFree
Knowledge cutoffopen
Output speed85 t/s
Statussolid
Family profile
Best published score in each covered benchmark family.
71/100 avg
- Reasoning
- 90
- Coding
- 49
- Agentic
- 64
- Long context
- 79
4 tested benchmark families
Benchmark placements
Per-benchmark positions
Where Hy3 places on each public benchmark source that publishes a row. Rank counts every model with a latest row in the same test — not a universal quality score.
| Benchmark | Family | Rank | Score | Source | Date |
|---|---|---|---|---|---|
| GPQA Diamond | Reasoning | #48/ 184 | 90 | Artificial Analysis | 2026-09-28 |
| Terminal-Bench | Agentic | #48/ 167 | 64 | Artificial Analysis | 2026-09-28 |
| SciCode | Coding | #53/ 97 | 49 | Artificial Analysis | 2026-09-28 |
| AA-LCR | Long context | #55/ 184 | 79 | Artificial Analysis | 2026-09-28 |
| Humanity's Last Exam | Reasoning | #70/ 186 | 34 | Artificial Analysis | 2026-09-28 |
| AA output speed | Performance | #75/ 195 | Artificial Analysis | 2026-10-12 |
Family context
Strongest rows in each family
The three highest-scoring models with pages in each capability family. Where Hy3 shows up, it's highlighted.
ReasoningHy3 · 90
- 196
- 295
- 395
Math
- 152
- 252
- 350
CodingHy3 · 49
- 174
- 273
- 370
AgenticHy3 · 64
- 191
- 289
- 389
Long contextHy3 · 79
- 189
- 286
- 385
Newest receipts
6 recentLast 120 days
Benchmark rows added to the public ledger for Hy3 in the last 120 days. Older rows live in the full table below.
| Benchmark | Family | Score | Source | Days ago |
|---|---|---|---|---|
| AA output speed | Performance | 85 t/s | Artificial Analysis | 10 |
| AA-LCR | Long context | 79% | Artificial Analysis | 24 |
| GPQA Diamond | Reasoning | 89.7% | Artificial Analysis | 24 |
| Humanity's Last Exam | Reasoning | 33.5% | Artificial Analysis | 24 |
| SciCode | Coding | 48.6% | Artificial Analysis | 24 |
| Terminal-Bench | Agentic | 64.42% | Artificial Analysis | 24 |
Family coverage
Best score per family
A high composite that hides a weak family is a trap. These bars surface the families where this model hasn't been publicly tested, and where it leads.
Price vs. performance
Where Hy3 sits on the scatter
Hover any dot for name, score, and input price. Keyboard: tab through the top twelve, or use the ranking below.
AnthropicOpenAIGoogleAlibabaMetaZhipuxAIDeepSeekMiniMaxXiaomiMoonshotOtherMeituanMistralNVIDIACohere
Full benchmark ledger
Every catalog benchmark for Hy3. Scores link to the original source; gaps mean no public row exists yet.
6 of 13 catalog benchmarks have a sourced row for Hy3.
Reasoning
Humanity's Last ExamVerified
33.5%
HLE (AA run)
Coding
SciCodeVerified
48.6%
SciCode (AA run)
Agentic
Terminal-BenchVerified
64.42%
Terminal-Bench v2.1 (AA run)
Long context
AA-LCRVerified
79%
AA-LCR (AA run)
Performance
AA output speedVerified
85 t/s
Median output tokens/s (1k prompt, default provider)
Sources
Where the numbers come from
- Artificial Analysis models index· checked 2026-07-21
Questions this model page answers
Every answer below is assembled from the dated fields on this page. Nothing is written separately for search.
What is Hy3?
Hy3 is a Other model: Tencent Hunyuan Hy3. Superseded as the current Hunyuan preview by Hy4 preview (28 Aug 2026, 770B / 49B, $0.834 / $2.501).
What is Hy3 for?
Frontier model tracked on Artificial Analysis. Use the benchmark ledger below for sourced scores; we do not infer numbers beyond published rows.
What does Hy3 cost?
Hy3 lists Free input and Free output per 1M tokens. Pricing last checked 2026-07-21.
How large is the Hy3 context window?
Hy3 publishes a 128K token context window.
What is the VerdictPal status of Hy3?
Hy3 carries Solid status on the model atlas.
The Pack · Editorial newsletter
Know when the rules move.
One short email when a university changes its AI rule for theses, or a research tool changes its pricing or privacy terms. No tracking, no third-party analytics. Unsubscribe in one click.