Benchmark menu

Model detail

Hy3 Preview

A static, shareable model page that remains complete without JavaScript.

Hy3 Preview

Provider: Tencent · Public slug: tencent-hy3-preview

Rank
24
Rank movement
Baseline
Score
28.6
Last measured
2026-07-04
Coverage
Medium not supported

Reasoning scores

XHigh
28.6
Medium
None
28.6

Generation health

XHigh56%22%22%n 9None50%25%25%n 8 Medium not supportedHigh failure rate

Cost summary

Mean standard list-price equivalent
$0.0023
Median unique-cell cost
$0.0015
Unique-cell range
$0.000909–$0.0075
Coverage
16 unique cells / 17 observed runs
Observed-run total
$0.04
Evidence basis
Provider-reported
Mean output
7.7k tok
Output range
1.4k–28k tok
Token samples
17

Cost means equal-weight unique model/game/reasoning cells. Provider-account billed spend is reported separately on Charts.

Run setup

  • Overall Models: combines public reasoning variants
  • Standard GameBench code-generation prompt