Compare · ModelsLive · 2 picked · head to head
DeepSeek V4 Pro vs Grok 4.3
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
DeepSeek V4 Pro wins on 19/23 benchmarks
DeepSeek V4 Pro wins 19 of 23 shared benchmarks. Leads in speed · arena · knowledge.
Category leads
speed·DeepSeek V4 Proarena·DeepSeek V4 Proknowledge·DeepSeek V4 Progeneral·DeepSeek V4 Promath·DeepSeek V4 Procoding·DeepSeek V4 Proreasoning·DeepSeek V4 Prolanguage·Grok 4.3
Hype vs Reality
Attention vs performance
DeepSeek V4 Pro
#96 by perf·#4 by attention
Grok 4.3
#76 by perf·#14 by attention
Best value
DeepSeek V4 Pro
1.4x better value than Grok 4.3
DeepSeek V4 Pro
41.0 pts/$
$1.32/M
Grok 4.3
30.0 pts/$
$1.88/M
Vendor risk
Mixed exposure
One or more vendors flagged
DeepSeek
$3.4B·Tier 1
xAI
$250.0B·Tier 1
Head to head
23 benchmarks · 2 models
DeepSeek V4 ProGrok 4.3
Artificial Analysis · Agentic Index
DeepSeek V4 Pro leads by +12.3
Artificial Analysis Agentic Index · a composite score measuring how well a model performs in agentic workflows · multi-step tool use, planning, error recovery, and autonomous task completion. Aggregates results from multiple agentic benchmarks including SWE-bench, tool-use tests, and planning evaluations. The canonical single-number metric for "how good is this model as an agent?"
DeepSeek V4 Pro
36.4
Grok 4.3
24.1
Artificial Analysis · Coding Index
DeepSeek V4 Pro leads by +17.1
Artificial Analysis Coding Index · a composite score that aggregates performance across multiple coding benchmarks into a single index. Tracks code generation quality, debugging ability, multi-language competence, and real-world software engineering tasks. Used by Artificial Analysis to rank model coding capability in a normalized, comparable format. Useful for developers choosing between models for coding-heavy workloads.
DeepSeek V4 Pro
59.4
Grok 4.3
42.3
Artificial Analysis · Quality Index
Grok 4.3 leads by +1.6
DeepSeek V4 Pro
36.0
Grok 4.3
37.6
Chatbot Arena Elo · Coding
DeepSeek V4 Pro leads by +89.3
DeepSeek V4 Pro
1446.0
Grok 4.3
1356.6
Chatbot Arena Elo · Overall
DeepSeek V4 Pro leads by +16.3
DeepSeek V4 Pro
1457.8
Grok 4.3
1441.4
Chess Puzzles
DeepSeek V4 Pro leads by +1.0
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
DeepSeek V4 Pro
14.8
Grok 4.3
13.7
Dtbench
DeepSeek V4 Pro
84.5
Grok 4.3
84.5
FrontierMath-Tier-4-v2-Private
Grok 4.3 leads by +12.2
DeepSeek V4 Pro
2.4
Grok 4.3
14.6
FrontierMath-Tiers-1-3-v2-Private
DeepSeek V4 Pro leads by +2.6
DeepSeek V4 Pro
47.6
Grok 4.3
45.1
GPQA diamond
DeepSeek V4 Pro leads by +2.9
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
DeepSeek V4 Pro
91.5
Grok 4.3
88.7
LiveBench · Agentic Coding
DeepSeek V4 Pro leads by +6.7
DeepSeek V4 Pro
56.7
Grok 4.3
50.0
LiveBench · Coding
DeepSeek V4 Pro leads by +0.1
DeepSeek V4 Pro
70.0
Grok 4.3
69.9
LiveBench · Data Analysis
DeepSeek V4 Pro leads by +18.8
DeepSeek V4 Pro
74.5
Grok 4.3
55.8
LiveBench · If
Grok 4.3 leads by +0.4
DeepSeek V4 Pro
62.4
Grok 4.3
62.8
LiveBench · Language
DeepSeek V4 Pro leads by +4.5
DeepSeek V4 Pro
78.1
Grok 4.3
73.6
LiveBench · Mathematics
DeepSeek V4 Pro leads by +6.3
DeepSeek V4 Pro
90.7
Grok 4.3
84.3
LiveBench · Overall
DeepSeek V4 Pro leads by +6.8
DeepSeek V4 Pro
73.6
Grok 4.3
66.7
LiveBench · Reasoning
DeepSeek V4 Pro leads by +11.9
DeepSeek V4 Pro
82.7
Grok 4.3
70.8
Lmca
DeepSeek V4 Pro leads by +3.4
DeepSeek V4 Pro
48.5
Grok 4.3
45.1
OTIS Mock AIME 2024-2025
DeepSeek V4 Pro leads by +3.3
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
DeepSeek V4 Pro
96.7
Grok 4.3
93.3
Proofbench
DeepSeek V4 Pro leads by +5.0
DeepSeek V4 Pro
16.0
Grok 4.3
11.0
SimpleQA Verified
DeepSeek V4 Pro leads by +13.8
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
DeepSeek V4 Pro
47.0
Grok 4.3
33.2
WeirdML
Grok 4.3 leads by +1.0
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
DeepSeek V4 Pro
48.9
Grok 4.3
49.9
Full benchmark table
| Benchmark | DeepSeek V4 Pro | Grok 4.3 |
|---|---|---|
Artificial Analysis · Agentic Index Artificial Analysis Agentic Index · a composite score measuring how well a model performs in agentic workflows · multi-step tool use, planning, error recovery, and autonomous task completion. Aggregates results from multiple agentic benchmarks including SWE-bench, tool-use tests, and planning evaluations. The canonical single-number metric for "how good is this model as an agent?" | 36.4 | 24.1 |
Artificial Analysis · Coding Index Artificial Analysis Coding Index · a composite score that aggregates performance across multiple coding benchmarks into a single index. Tracks code generation quality, debugging ability, multi-language competence, and real-world software engineering tasks. Used by Artificial Analysis to rank model coding capability in a normalized, comparable format. Useful for developers choosing between models for coding-heavy workloads. | 59.4 | 42.3 |
Artificial Analysis · Quality Index | 36.0 | 37.6 |
Chatbot Arena Elo · Coding | 1446.0 | 1356.6 |
Chatbot Arena Elo · Overall | 1457.8 | 1441.4 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 14.8 | 13.7 |
Dtbench | 84.5 | 84.5 |
FrontierMath-Tier-4-v2-Private | 2.4 | 14.6 |
FrontierMath-Tiers-1-3-v2-Private | 47.6 | 45.1 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 91.5 | 88.7 |
LiveBench · Agentic Coding | 56.7 | 50.0 |
LiveBench · Coding | 70.0 | 69.9 |
LiveBench · Data Analysis | 74.5 | 55.8 |
LiveBench · If | 62.4 | 62.8 |
LiveBench · Language | 78.1 | 73.6 |
LiveBench · Mathematics | 90.7 | 84.3 |
LiveBench · Overall | 73.6 | 66.7 |
LiveBench · Reasoning | 82.7 | 70.8 |
Lmca | 48.5 | 45.1 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 96.7 | 93.3 |
Proofbench | 16.0 | 11.0 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 47.0 | 33.2 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 48.9 | 49.9 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.88 | $1.76 | 1.0M tokens (~524 books) | $11.01 | |
| $1.25 | $2.50 | 1.0M tokens (~500 books) | $15.63 |
People also compared