Compare · ModelsLive · 2 picked · head to head
DeepSeek V4 Pro 0813 vs Grok 4.6
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Grok 4.6 wins on 12/16 benchmarks
Grok 4.6 wins 12 of 16 shared benchmarks. Leads in agentic · general · coding.
Category leads
agentic·Grok 4.6reasoning·DeepSeek V4 Pro 0813knowledge·DeepSeek V4 Pro 0813general·Grok 4.6coding·Grok 4.6math·Grok 4.6
Hype vs Reality
Attention vs performance
DeepSeek V4 Pro 0813
#57 by perf·#4 by attention
Grok 4.6
#74 by perf·#14 by attention
Best value
DeepSeek V4 Pro 0813
1.8x better value than Grok 4.6
DeepSeek V4 Pro 0813
25.8 pts/$
$2.28/M
Grok 4.6
14.2 pts/$
$4.00/M
Vendor risk
Mixed exposure
One or more vendors flagged
DeepSeek
$3.4B·Tier 1
xAI
$250.0B·Tier 1
Head to head
16 benchmarks · 2 models
DeepSeek V4 Pro 0813Grok 4.6
APEX-Agents
Grok 4.6 leads by +18.0
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
DeepSeek V4 Pro 0813
47.3
Grok 4.6
65.3
ARC-AGI
DeepSeek V4 Pro 0813 leads by +3.0
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
DeepSeek V4 Pro 0813
90.5
Grok 4.6
87.5
ARC-AGI-2
Grok 4.6 leads by +5.8
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
DeepSeek V4 Pro 0813
61.3
Grok 4.6
67.1
Chess Puzzles
DeepSeek V4 Pro 0813 leads by +13.7
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
DeepSeek V4 Pro 0813
43.2
Grok 4.6
29.5
Dtbench
Grok 4.6 leads by +5.8
DeepSeek V4 Pro 0813
89.8
Grok 4.6
95.5
Frontiercode
Grok 4.6 leads by +19.5
DeepSeek V4 Pro 0813
28.6
Grok 4.6
48.0
FrontierMath-Tier-4-v2-Private
Grok 4.6 leads by +4.9
DeepSeek V4 Pro 0813
26.8
Grok 4.6
31.7
FrontierMath-Tiers-1-3-v2-Private
Grok 4.6 leads by +1.5
DeepSeek V4 Pro 0813
68.0
Grok 4.6
69.4
GPQA diamond
Grok 4.6 leads by +3.3
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 0813
92.6
Grok 4.6
95.8
Lmca
Grok 4.6 leads by +3.5
DeepSeek V4 Pro 0813
53.5
Grok 4.6
57.0
Mirrorcode
Grok 4.6 leads by +7.8
DeepSeek V4 Pro 0813
33.3
Grok 4.6
41.1
Mystery Game Puzzles
DeepSeek V4 Pro 0813 leads by +9.9
DeepSeek V4 Pro 0813
37.2
Grok 4.6
27.3
OTIS Mock AIME 2024-2025
Grok 4.6 leads by +0.6
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
DeepSeek V4 Pro 0813
98.6
Grok 4.6
99.2
Proofbench
Grok 4.6 leads by +1.0
DeepSeek V4 Pro 0813
50.0
Grok 4.6
51.0
SimpleQA Verified
DeepSeek V4 Pro 0813 leads by +3.6
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
DeepSeek V4 Pro 0813
52.9
Grok 4.6
49.3
WeirdML
Grok 4.6 leads by +1.1
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
DeepSeek V4 Pro 0813
66.2
Grok 4.6
67.3
Full benchmark table
| Benchmark | DeepSeek V4 Pro 0813 | Grok 4.6 |
|---|---|---|
APEX-Agents APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments. | 47.3 | 65.3 |
ARC-AGI ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization. | 90.5 | 87.5 |
ARC-AGI-2 ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data. | 61.3 | 67.1 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 43.2 | 29.5 |
Dtbench | 89.8 | 95.5 |
Frontiercode | 28.6 | 48.0 |
FrontierMath-Tier-4-v2-Private | 26.8 | 31.7 |
FrontierMath-Tiers-1-3-v2-Private | 68.0 | 69.4 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 92.6 | 95.8 |
Lmca | 53.5 | 57.0 |
Mirrorcode | 33.3 | 41.1 |
Mystery Game Puzzles | 37.2 | 27.3 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 98.6 | 99.2 |
Proofbench | 50.0 | 51.0 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 52.9 | 49.3 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 66.2 | 67.3 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.11 | $4.44 | 1.0M tokens (~524 books) | $11.93 | |
| $2.00 | $6.00 | 500K tokens (~250 books) | $30.00 |