The AI arena is free today

Open Superagent

Hy3 vs Hy4 preview

Hy4 preview leads the LLM Stats Score 51.5 to 43.1.

Tencent · Tencent · Updated for 2026

Which is better?

Hy4 preview leads the overall LLM Stats Score 51.5 to 43.1, ranking #10 overall.

In the 17 individual benchmarks reported for both models, Hy4 preview wins 16; this is a narrower head-to-head signal than the composite indexes.

Based on current LLM Stats indexes, shared benchmarks, pricing, and model metadata for 2026.

Choose Hy3

  • you are already invested in the Tencent ecosystem

Choose Hy4 preview

  • overall performance matters — it scores 51.5 and ranks #10 on LLM Stats
  • your work emphasizes reasoning and coding — it leads those capability indexes
  • you value its reported benchmark strengths — it wins 16 of 17 exact shared results
  • you want the most recent training data — it shipped Aug 2026

At a glance

The differences that matter most.

Hy3
Tencent
Hy4 preview
Tencent
Core performance indexes
LLM Stats Score
43.1
#38
51.5
#10
Reasoning
43.1
#37
51.2
#8
Coding
33.4
#29
40.9
#10
Agents
26.7
#39
37.1
#11
Cost, coverage & limits
Benchmark wins
1 of 17
16 of 17
Input price
— / M
— / M
Output price
— / M
— / M
Context window
—
—

Capability indexes

Additional strengths measured across groups of related public benchmarks

2 shared
Index
Hy3
Hy4 preview
Math
36.3#38
40.3#12
Tool use
22.9#42
32.6#7
Conservative TrueSkill rating · higher is betterHow scores work

Individual benchmarks

31 reported for Hy3 · 32 for Hy4 preview

17 shared

Hy3 outperforms in 1 benchmarks (Humanity's Last Exam (no tools, text-only)), while Hy4 preview is better at 16 benchmarks (APEX-Agents, ArXivMath, DeepSWE, GPQA, HorizonMath, Humanity's Last Exam (with tools, text-only), MathArena Apex, MCP Atlas, NL2Repo, SkillsBench, SuperChem, SWE-bench Multilingual, SWE-Bench Pro, Terminal-Bench 2.1, Toolathlon, WideSearch).

Hy4 preview significantly outperforms across most benchmarks.

Mon Aug 31 2026 • llm-stats.com

Human preference

Blind head-to-head votes and playground preference scores

Model Size

Parameter count comparison

475.0B diff

Hy4 preview has 475.0B more parameters than Hy3, making it 161.0% larger.

Tencent
Hy3
295.0Bparameters
Tencent
Hy4 preview
770.0Bparameters
295.0B
Hy3
770.0B
Hy4 preview

License

Usage and distribution terms

Both models are licensed under Apache 2.0.

Both models share the same licensing terms, providing consistent usage rights.

Hy3

Apache 2.0

Open weights

Hy4 preview

Apache 2.0

Open weights

Release Timeline

When each model was launched

Hy3 was released on 2026-07-06, while Hy4 preview was released on 2026-08-28.

Hy4 preview is 2 months newer than Hy3.

Hy3

Jul 6, 2026

1 months ago

Hy4 preview

Aug 28, 2026

3 days ago

1mo newer

Knowledge Cutoff

When training data ends

Neither model specifies a knowledge cutoff date.

Unable to compare the recency of their training data.

No cutoff dates available

Outputs Comparison

Notice missing or incorrect data?Start an Issue discussion→

Judge for yourself.

Run your own prompts against Hy3 and Hy4 preview side-by-side, then vote on the output you prefer.

Hy3
✓ Preferred
Hy4 preview
Open in Playground

FAQ

Common questions about Hy3 vs Hy4 preview.

Which is better, Hy3 or Hy4 preview?

Hy4 preview leads the LLM Stats Score 51.5 to 43.1. Hy3 is made by Tencent and Hy4 preview is made by Tencent. The best choice depends on your use case — compare their capability indexes, individual benchmarks, pricing, and limits above.

How does Hy3 compare to Hy4 preview in benchmarks?

Hy3 scores DeepSearchQA: 91.0%, GPQA: 90.4%, IMO-AnswerBench: 90.0%, BrowseComp: 84.2%, MCP Atlas: 79.1%. Hy4 preview scores GPQA: 92.3%, Terminal-Bench 2.1: 85.4%, WideSearch: 83.9%, MCP Atlas: 83.7%, SWE-bench Multilingual: 82.9%.

What are the main differences between Hy3 and Hy4 preview?

Key differences include LLM Stats Score (43.1 vs 51.5). See the full comparison above for benchmark-by-benchmark results.

Related comparisons