Noometry

Model comparison

DeepSeek-R1 vs Grok 4

Grok 4 is the stronger model overall, scoring 48.1 to 42.3 on the Noometry Index.

Last verified . 38 shared benchmarks.

DeepSeek-R1 DeepSeek

42.3

Rank #115 Confirmed

Grok 4 xAI

48.1

Rank #56 Confirmed

Summary

  • They share 38 benchmarks with published results for both. DeepSeek-R1 scores higher in 2 categories and Grok 4 in 7 categories; 8 gaps are clear of the uncertainty.
  • The widest gap is in reasoning, where Grok 4 leads 36.7 to 18.6.
  • The biggest single-benchmark swing is ARC-AGI-1: 21.2% for DeepSeek-R1 and 66.7% for Grok 4.

Side by side

DeepSeek-R1Grok 4
ProviderDeepSeekxAI
Noometry Index42.348.1
Released2025-01-202025-07-09
WeightsProprietaryProprietary
Context window164K—
Max output64K—
Input $ / M tokens$0.50—
Output $ / M tokens$2.15—
Results tracked5248

Sponsored placements are available on pages like this one. Advertise on Noometry

Category by category

Coding Grok 4 leads

DeepSeek-R1: 46.3 (#68), Grok 4: 50.3 (#46)

BenchmarkDeepSeek-R1Grok 4
Aider Polyglot71.4%79.6%
WeirdML41.6%45.7%
LMArena Coding14271408
SciCode35.7%—
LiveBench Coding66.7%—
ALE-Bench804.12—
AlgoTune1.7—

Agentic & Tool Use Grok 4 leads

DeepSeek-R1: 30.7 (#75), Grok 4: 32.3 (#68)

BenchmarkDeepSeek-R1Grok 4
DeepResearch Bench35.1%47.3%
BALROG34.9%43.6%
METR Time Horizons53.8%66.6%
Terminal-Bench—27.2%
Berkeley Function Calling Leaderboard—63%
GDPval—21.1%
Cybench—43%
LMArena Search—1142

Reasoning Grok 4 leads

DeepSeek-R1: 18.6 (#278), Grok 4: 36.7 (#65)

BenchmarkDeepSeek-R1Grok 4
ARC-AGI-21.3%16%
SimpleBench40.8%60.5%
Kagi LLM Benchmark69.4%73.6%
ARC-AGI-121.2%66.7%
LMArena Hard Prompts14161409
Epoch Capabilities Index141.29146.44
ForecastBench6060.9
CritPt1.1%—
Chess Puzzles—28%
LiveBench Reasoning83.2%—
LiveBench Data Analysis69.8%—
LiveBench71.6%—

Math Grok 4 leads

DeepSeek-R1: 43.8 (#79), Grok 4: 48.4 (#64)

BenchmarkDeepSeek-R1Grok 4
OTIS Mock AIME 2024-202566.4%84%
Omni-MATH42.4%60.3%
LMArena Math14001422
LiveBench Math80.7%—
MATH Level 596.6%—
FrontierMath (Feb 2025 set)—19.7%
FrontierMath Tier 4 (v1)—2.1%

Knowledge Grok 4 leads

DeepSeek-R1: 44.5 (#87), Grok 4: 53.8 (#55)

BenchmarkDeepSeek-R1Grok 4
GPQA Diamond76.3%87%
MMLU-Pro79.3%85.1%
Confabulations12.7%12.4%
GPQA (HELM)66.6%72.7%
LMArena Expert13941415
Vectara Hallucination Rate11.3%—

Multimodal Not comparable

DeepSeek-R1: —, Grok 4: 33.7 (#94)

BenchmarkDeepSeek-R1Grok 4
LMArena Vision—1210
GeoBench—45%

Multilingual Too close to call

DeepSeek-R1: 52.4 (#85), Grok 4: 51.8 (#103)

BenchmarkDeepSeek-R1Grok 4
LMArena Non-English14121403
LMArena Chinese14421427
LMArena French14171418
LMArena German14041429
LMArena Japanese13911394
LMArena Korean13601377
LMArena Russian14231410
LMArena Spanish14111420

Instruction Following Grok 4 leads

DeepSeek-R1: 72.0 (#143), Grok 4: 79.2 (#5)

BenchmarkDeepSeek-R1Grok 4
IFEval78.4%94.9%
LMArena Instruction Following13821387
LiveBench Instruction Following80.5%—

Long Context Grok 4 leads

DeepSeek-R1: 45.4 (#36), Grok 4: 63.1 (#4)

BenchmarkDeepSeek-R1Grok 4
Fiction.LiveBench75%94.4%
LMArena Longer Query13911409

Writing & Preference DeepSeek-R1 leads

DeepSeek-R1: 61.4 (#88), Grok 4: 58.5 (#116)

BenchmarkDeepSeek-R1Grok 4
LMArena Text14281411
LMArena Creative Writing14051397
Short-Story Creative Writing83%76.9%
WildBench82.8%79.7%
LMArena Multi-Turn14051416
EQ-Bench Creative Writing1500—
LiveBench Language48.5%—

Frequently asked questions

Is DeepSeek-R1 better than Grok 4?

Grok 4 is the stronger model overall, scoring 48.1 to 42.3 on the Noometry Index.

Is DeepSeek-R1 or Grok 4 better for coding?

Grok 4 scores higher on coding benchmarks: 50.3 versus 46.3 in the Noometry coding category.

How many benchmarks do DeepSeek-R1 and Grok 4 share?

38 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and Grok 4 has 48.

Related comparisons

Go deeper