Tencent
Hy3
Tencent Hy reasoning model for coding, instruction following, and agent tasks
tencent/hy3
Context
256K
Output
128K
Released
2026-07-06
Knowledge cutoff
—
Types
Reasoning
Tool use
Structured output
Attachments
Open weights
Provider pricing
- Price
- $0.07 / $0.26
- Context
- 262.1K
- Status
- Stable
- Price
- $0.08 / $0.33
- Context
- 262.1K
- Status
- Stable
- Price
- $0.08 / $0.33
- Context
- 262.1K
- Status
- Stable
- Price
- $0.13 / $0.53
- Context
- 256K
- Status
- Stable
- Price
- $0.13 / $0.53
- Context
- 262.1K
- Status
- Stable
- Price
- $0.14 / $0.58
- Context
- 262.1K
- Status
- Stable
- Price
- $0.14 / $0.58
- Context
- 262.1K
- Status
- Stable
- Price
- $0.16 / $0.64
- Context
- 262.1K
- Status
- Stable
- Price
- $0.18 / $0.59
- Context
- 256K
- Status
- Stable
- Price
- $0.28 / $1.10
- Context
- 262.1K
- Status
- Stable
Benchmarks
SWE-Bench Verified
78 · resolved
SWE-Bench Multilingual
75.8 · score
SWE-Bench Pro
57.9 · score
Terminal-Bench
71.7 · score
NL2Repo
45.6 · score
DeepSWE
28 · score
BrowseComp
84.2 · score
WideSearch
76.4 · score
DeepSearchQA
91 · score
MCP Atlas
79.1 · score
Toolathlon
48.5 · score
APEX-Agents
25.6 · pass@1
ClawEval
68.5 · pass@3
WildClawBench
53.6 · score
SkillsBench
55.3 · average over 3 runs
Humanity's Last Exam
53.2 · score
Humanity's Last Exam
37 · score
GPQA Diamond
90.4 · score
FrontierScience
21.3 · score
FrontierScience
74.8 · score
USAMO
72 · score
MathArena Apex
38.7 · score
ArxivMath
52.2 · score
HorizonMath
7.1 · pass@12
PHYBench
77.4 · score
CMT Benchmark
37.8 · score
IMOAnswerBench
90 · score
SuperChem
54.9 · score
CL-bench
23.8 · score
CL-bench-life
17 · score
AA-LCR
73.4 · score