readable AI benchmarks (simplified)

readable AI benchmarks

Tencent

Hy3-preview

non-reasoning · open weights · Apr 23, 2026

Hy3-preview has a Quality Score of 24.2, ranking 52nd among 53 scored models. Per 1M tokens, pricing is $0.065 input, $0.235 output, and $0.025 cached input. Its Reliability Score is 31.8, ranking 51st among 53 scored models, below average (total average: 50.1). Its Value Score is 29.2, low compared with the total Value average of 39.6.

Quality Score24.2coverage 78.8%: missing Epoch General ECI, Coding, CursorBench 3.2, Agentic and DeepSWE v1.1
Value Score29.2
Reliability31.8
Cache Discount61%

Model specification

Reasoning
non-reasoning
Input modalities
none
Output modalities
none
Context window
not available
Weights
open weights

Token prices USD per 1M tokens

Input
$0.065
Output
$0.235
Cached Input
$0.025
Cache write
not available
Cost per task
not available

Capability

Intelligence
26.1
Coding
not available
Agentic
not available
Omniscience
-36.4
Correct
22.4%
Blended price
$0.054

Answer outcomes

Attempt rate: 81.1%

Correct 22.4%Incorrect 58.8%Abstained 18.9%

Artificial Analysis benchmarks

SciCode
39%
Humanity’s Last Exam
6%
GPQA Diamond
73%
CritPt
0%
AA-Omniscience
32%
AA-LCR
34%

Similar models