Compare · ModelsLive · 2 picked · head to head
Claude Sonnet 5 vs DeepSeek V4 Pro 0813
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Claude Sonnet 5 wins on 7/13 benchmarks
Claude Sonnet 5 wins 7 of 13 shared benchmarks. Leads in agentic · coding · math.
Category leads
agentic·Claude Sonnet 5knowledge·DeepSeek V4 Pro 0813general·DeepSeek V4 Pro 0813coding·Claude Sonnet 5math·Claude Sonnet 5
Hype vs Reality
Attention vs performance
Claude Sonnet 5
#66 by perf·#10 by attention
DeepSeek V4 Pro 0813
#47 by perf·#6 by attention
Best value
DeepSeek V4 Pro 0813
2.3x better value than Claude Sonnet 5
Claude Sonnet 5
9.5 pts/$
$6.00/M
DeepSeek V4 Pro 0813
21.6 pts/$
$2.77/M
Vendor risk
Mixed exposure
One or more vendors flagged
Anthropic
$965.0B·Tier 1
DeepSeek
$3.4B·Tier 1
Head to head
13 benchmarks · 2 models
Claude Sonnet 5DeepSeek V4 Pro 0813
APEX-Agents
Claude Sonnet 5 leads by +7.2
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
Claude Sonnet 5
54.5
DeepSeek V4 Pro 0813
47.3
Chess Puzzles
DeepSeek V4 Pro 0813 leads by +12.6
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
Claude Sonnet 5
31.6
DeepSeek V4 Pro 0813
44.2
Dtbench
DeepSeek V4 Pro 0813 leads by +2.3
Claude Sonnet 5
87.5
DeepSeek V4 Pro 0813
89.8
Frontiercode
Claude Sonnet 5 leads by +14.2
Claude Sonnet 5
42.7
DeepSeek V4 Pro 0813
28.6
FrontierMath-Tier-4-v2-Private
Claude Sonnet 5 leads by +2.4
Claude Sonnet 5
29.3
DeepSeek V4 Pro 0813
26.8
FrontierMath-Tiers-1-3-v2-Private
Claude Sonnet 5 leads by +1.0
Claude Sonnet 5
65.6
DeepSeek V4 Pro 0813
64.6
GPQA diamond
DeepSeek V4 Pro 0813 leads by +1.5
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Claude Sonnet 5
87.4
DeepSeek V4 Pro 0813
88.9
Lmca
Claude Sonnet 5 leads by +5.4
Claude Sonnet 5
58.9
DeepSeek V4 Pro 0813
53.5
Mystery Game Puzzles
DeepSeek V4 Pro 0813 leads by +8.8
Claude Sonnet 5
28.4
DeepSeek V4 Pro 0813
37.2
OTIS Mock AIME 2024-2025
DeepSeek V4 Pro 0813 leads by +3.9
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Claude Sonnet 5
94.7
DeepSeek V4 Pro 0813
98.6
Proofbench
Claude Sonnet 5 leads by +27.0
Claude Sonnet 5
77.0
DeepSeek V4 Pro 0813
50.0
SimpleQA Verified
DeepSeek V4 Pro 0813 leads by +19.2
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
Claude Sonnet 5
33.7
DeepSeek V4 Pro 0813
52.9
WeirdML
Claude Sonnet 5 leads by +2.6
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
Claude Sonnet 5
68.8
DeepSeek V4 Pro 0813
66.2
Full benchmark table
| Benchmark | Claude Sonnet 5 | DeepSeek V4 Pro 0813 |
|---|---|---|
APEX-Agents APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments. | 54.5 | 47.3 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 31.6 | 44.2 |
Dtbench | 87.5 | 89.8 |
Frontiercode | 42.7 | 28.6 |
FrontierMath-Tier-4-v2-Private | 29.3 | 26.8 |
FrontierMath-Tiers-1-3-v2-Private | 65.6 | 64.6 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 87.4 | 88.9 |
Lmca | 58.9 | 53.5 |
Mystery Game Puzzles | 28.4 | 37.2 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 94.7 | 98.6 |
Proofbench | 77.0 | 50.0 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 33.7 | 52.9 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 68.8 | 66.2 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $2.00 | $10.00 | 1.0M tokens (~500 books) | $40.00 | |
| $0.55 | $5.00 | 1.0M tokens (~524 books) | $16.63 |