AI BENCHY
Advertise here
#275

Hy4 preview

Tencent Release: 2026-08-29 Tested on: 2026-08-29 14:22 tencent/hy4-preview::none
770B total (49B active)MoEOpen source
(low) (none)

Summary

Hy4 preview scores 4.1 on AI BENCHY and ranks #275. It has 1.3 reliability, a 27.3% pass rate, $0.035 total cost, and 51.97s average response time.

What makes Hy4 preview unique: It stands out most in Domain specific, where it ranks #1, while Anti-AI Tricks is its weakest area at #9.

Model facts

Researched on 2026-08-29

Reported
Parameters
770B total (49B active)
Architecture
MoE
Availability
Open source
License
Apache-2.0

Vendor-reported backbone counts exclude the native 10B total / 0.7B active MTP layer used for speculative decoding.

Parameter source 1 Parameter source 2 Availability source

Score

4.1

Consistency

7.8

Reliability

1.3

Total Cost (Current Price)

$0.035

Total Output Tokens

6,154

Total Input Tokens

23,473

Input Price

$0.834 / 1M

Output Price

$2.501 / 1M

Tests Correct

Wrong Tests: 19

Attempt pass rate: 27.3%

Flaky tests

6

Flaky tests had mixed outcomes across runs (at least one pass and one fail).

Response Time (avg)

51.97s

Response Time (max): 96.84s

Response Time (total): 1143.28s

API error: 10 Wrong answer: 6 Did not follow instructions: 2 Extra formatting: 1

Hamster playing table tennis

Prompt: Create a detailed SVG illustration of a hamster playing table tennis.

#275 Hy4 preview

none
This operation was aborted
Cost
$0.000
Time
0.0s
Tokens
0 tok

Run history

Tested on Score Reliability Tests Correct Total Cost Compare
2026-08-29 15:55 Re-test 4.7 3.0 $0.040 Compare
2026-08-29 14:22 Initial run 4.1 1.3 $0.035 Current run

Charts

Choose the first model, then click a second model to open a side-by-side page.

Total Output Tokens

Score vs Total Output Tokens

Quick Compare

Hy4 previewnonevsNemotron 3 SupernoneFree AvailableHy4 previewnonevsRing-2.6-1TnoneHy4 previewnonevsLing-2.6-flashnoneHy4 previewnonevsNemotron 3.5 LightninglowFree AvailableHy4 previewnonevsGPT-4o-mininoneHy4 previewnonevsGPT-5.4 NanononeHy4 previewnonevsDeepSeek V3.2noneHy4 previewnonevsTrinity Large ThinkinghighHy4 previewnonevsLaguna S 2.1lowFree AvailableHy4 previewnonevsKAT-Coder-Air V2.5none

Category Breakdown

Category Score Consistency Tests Correct
Anti-AI Tricks 3.1 9.9
Coding 3.1 7.2
Combined 3.0 10.0
Data parsing and extraction 6.3 5.8
Domain specific 5.9 7.2
General Intelligence 5.1 10.0
Instructions following 7.1 5.8
Puzzle Solving 4.0 4.9
Tool Calling 3.0 10.0
Trivia 3.0 10.0

Compared models