How do Hy3 and DeepSeek V4 Pro (Max) compare on coding benchmarks?

The comparison table shows SWE-bench, SWE-Pro, SWE-Atlas, Terminal-Bench, DeepSWE, and coding agent index scores for both Hy3 and DeepSeek V4 Pro (Max) when publicly disclosed.

LLM Comparison

Hy3 vs DeepSeek V4 Pro (Max): benchmark scores, pricing & comparison.

Q: Which is better: Hy3 or DeepSeek V4 Pro (Max)?

AskClash compares Hy3 and DeepSeek V4 Pro (Max) side by side across SWE-bench, GPQA, HLE, Terminal-Bench, coding agent scores, token pricing, and context window so you can see which model wins on each benchmark.

Side-by-side Hy3 vs DeepSeek V4 Pro (Max) comparison across SWE-bench, GPQA, HLE, Terminal-Bench, coding agent scores, token pricing, context window, and AskClash RWT. Green marks the winner on each benchmark.

Open live leaderboard Hy3 model page DeepSeek V4 Pro (Max) model page

Rank #17 vs #27AskClash overall scores 58.7 vs 48.7.

Pricing $0.12/$0.43 vs $0.43/$0.87Input and output token prices per 1M tokens when published.

Open Weight vs Open WeightTencent vs DeepSeek.

Hy3 vs DeepSeek V4 Pro (Max) benchmark comparison

Green cells highlight the winning model for each metric. Scores are cached from the AskClash LLM leaderboard snapshot.

Metric	Hy3	DeepSeek V4 Pro (Max)
Overall Score	58.7	48.7
Leaderboard Rank	#17	#27
RWT	8.0	7.0
HLE	53.2	37.7
GPQA	90.4	90.1
MATH-500	—	64.5
SWE-bench	78.0	80.6
SWE-Pro	57.9	55.4
Terminal-Bench	71.7	67.9
MCP Atlas	79.1	73.6
Tau2	92.7	96.2
MRCR	75.1	83.5
Input Price (per 1M tokens)	$0.12	$0.43
Output Price (per 1M tokens)	$0.43	$0.87
Context Window	256K	1M
Benchmarks Published	9	9

Hy3 vs DeepSeek V4 Pro (Max) head-to-head charts

Hy3 leads 7 and DeepSeek V4 Pro (Max) leads 3 of 10 shared benchmarks. Hy3 is cheaper on a defined 1M-input / 200K-output workload. Charts show only benchmarks both models publish.

Hy3DeepSeek V4 Pro (Max)

Overall

58.7Hy3

48.7DeepSeek V4 Pro (Max)

RWT

8.0Hy3

7.0DeepSeek V4 Pro (Max)

HLE

53.2Hy3

37.7DeepSeek V4 Pro (Max)

GPQA

90.4Hy3

90.1DeepSeek V4 Pro (Max)

SWE-bench

78.0Hy3

80.6DeepSeek V4 Pro (Max)

SWE-Pro

57.9Hy3

55.4DeepSeek V4 Pro (Max)

Terminal-Bench

71.7Hy3

67.9DeepSeek V4 Pro (Max)

MCP Atlas

79.1Hy3

73.6DeepSeek V4 Pro (Max)

Tau2

92.7Hy3

96.2DeepSeek V4 Pro (Max)

MRCR

75.1Hy3

83.5DeepSeek V4 Pro (Max)

Hy3

Input$0.12

Output$0.43

Workload$0.21

Context256K

DeepSeek V4 Pro (Max)

Input$0.43

Output$0.87

Workload$0.61

Context1M

Workload = published cost of 1M input + 200K output tokens. Open the live leaderboard for interactive compare charts.

More Hy3 and DeepSeek V4 Pro (Max) comparisons

Explore how Hy3 and DeepSeek V4 Pro (Max) stack up against other top-ranked LLMs.

Claude Fable 5 vs Hy3 Claude Fable 5 vs DeepSeek V4 Pro (Max)GPT-5.6 Sol vs Hy3 GPT-5.6 Sol vs DeepSeek V4 Pro (Max)Claude Opus 4.8 vs Hy3 Claude Opus 4.8 vs DeepSeek V4 Pro (Max)Grok 4.5 vs Hy3 Grok 4.5 vs DeepSeek V4 Pro (Max)GPT-5.5 vs Hy3 GPT-5.5 vs DeepSeek V4 Pro (Max)

How to read this comparison

Benchmark scores

Higher is better for all benchmark scores (SWE-bench, GPQA, HLE, Terminal-Bench, etc.). Green marks the model with the higher score.

Token pricing

Lower is better for input and output prices. Green marks the cheaper model per 1M tokens.

Coverage matters

Models with fewer disclosed benchmark cells may have inflated percentile scores. Check the benchmark cell count for context.

This comparison page is generated from the AskClash LLM leaderboard cache. Open the live leaderboard for real-time scores and interactive filtering.