How do Claude Sonnet 5 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 Claude Sonnet 5 and DeepSeek V4 Pro (Max) when publicly disclosed.

LLM Comparison

Claude Sonnet 5 vs DeepSeek V4 Pro (Max): benchmark scores, pricing & comparison.

Q: Which is better: Claude Sonnet 5 or DeepSeek V4 Pro (Max)?

AskClash compares Claude Sonnet 5 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 Claude Sonnet 5 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 Claude Sonnet 5 model page DeepSeek V4 Pro (Max) model page

Rank #8 vs #27AskClash overall scores 77.0 vs 48.7.

Pricing $3.00/$15.0 vs $0.43/$0.87Input and output token prices per 1M tokens when published.

Proprietary vs Open WeightAnthropic vs DeepSeek.

Claude Sonnet 5 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	Claude Sonnet 5	DeepSeek V4 Pro (Max)
Overall Score	77.0	48.7
Leaderboard Rank	#8	#27
RWT	8.5	7.0
HLE	57.4	37.7
GPQA	91.1	90.1
MATH-500	—	64.5
SWE-bench	85.2	80.6
SWE-Pro	63.2	55.4
Terminal-Bench	80.4	67.9
DeepSWE	53.8	—
OSWorld	81.2	—
MCP Atlas	—	73.6
CharXiv	88.3	—
MMMU-Pro	77.3	—
Tau2	—	96.2
MRCR	—	83.5
Input Price (per 1M tokens)	$3.00	$0.43
Output Price (per 1M tokens)	$15.0	$0.87
Context Window	1M	1M
Benchmarks Published	11	9

Claude Sonnet 5 vs DeepSeek V4 Pro (Max) head-to-head charts

Claude Sonnet 5 leads 7 and DeepSeek V4 Pro (Max) leads 0 of 7 shared benchmarks. DeepSeek V4 Pro (Max) is cheaper on a defined 1M-input / 200K-output workload. Charts show only benchmarks both models publish.

Claude Sonnet 5DeepSeek V4 Pro (Max)

Overall

77.0Claude Sonnet 5

48.7DeepSeek V4 Pro (Max)

RWT

8.5Claude Sonnet 5

7.0DeepSeek V4 Pro (Max)

HLE

57.4Claude Sonnet 5

37.7DeepSeek V4 Pro (Max)

GPQA

91.1Claude Sonnet 5

90.1DeepSeek V4 Pro (Max)

SWE-bench

85.2Claude Sonnet 5

80.6DeepSeek V4 Pro (Max)

SWE-Pro

63.2Claude Sonnet 5

55.4DeepSeek V4 Pro (Max)

Terminal-Bench

80.4Claude Sonnet 5

67.9DeepSeek V4 Pro (Max)

Claude Sonnet 5

Input$3.00

Output$15.0

Workload$6.00

Context1M

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 Claude Sonnet 5 and DeepSeek V4 Pro (Max) comparisons

Explore how Claude Sonnet 5 and DeepSeek V4 Pro (Max) stack up against other top-ranked LLMs.

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