Noometry

Model comparison

Grok 4.5 vs Qwen3-4B

Grok 4.5 is the stronger model overall, scoring 55.0 to 31.9 on the Noometry Index.

Last verified . 3 shared benchmarks.

Grok 4.5 xAI

55.0

Rank #25 Confirmed

Qwen3-4B Alibaba (Qwen)

31.9

Rank #264 Confirmed

Summary

  • They share 3 benchmarks with published results for both. Grok 4.5 scores higher in 4 categories and Qwen3-4B in 0 categories; 4 gaps are clear of the uncertainty.
  • The widest gap is in reasoning, where Grok 4.5 leads 56.1 to 19.2.
  • The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 97.8% for Grok 4.5 and 52.2% for Qwen3-4B.
  • Qwen3-4B has downloadable open weights; the other is API-only.

Side by side

Grok 4.5Qwen3-4B
ProviderxAIAlibaba (Qwen)
Noometry Index55.031.9
Released2026-07-082025-04-29
WeightsProprietaryOpen
Context window500K—
Max output500K—
Input $ / M tokens$2—
Output $ / M tokens$6—
Results tracked526

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

Category by category

Coding Not comparable

Grok 4.5: 52.2 (#35), Qwen3-4B: —

BenchmarkGrok 4.5Qwen3-4B
DeepSWE53.8%—
FrontierCode42.4%—
LMArena WebDev1553—
SciCode54.1%—
WeirdML46.4%—
LMArena Coding1474—
ALE-Bench1,309—

Agentic & Tool Use Grok 4.5 leads

Grok 4.5: 44.4 (#17), Qwen3-4B: 27.6 (#100)

BenchmarkGrok 4.5Qwen3-4B
APEX-Agents56.2%—
Berkeley Function Calling Leaderboard—35.7%
τ²-bench Banking47.9%—
PostTrainBench23.4%—
GBAEval65.4%—
GDP.pdf14%—
LMArena Search1213—
Vending-Bench 23,887—

Reasoning Grok 4.5 leads

Grok 4.5: 56.1 (#25), Qwen3-4B: 19.2 (#268)

BenchmarkGrok 4.5Qwen3-4B
Chess Puzzles36%4%
ARC-AGI-252.6%—
SimpleBench70%—
Kagi LLM Benchmark83.5%—
NYT Connections (extended)79.9%—
ARC-AGI-187.2%—
CritPt15.4%—
LMArena Hard Prompts1462—
DTBench96.5%—
LMCA45.2%—
Surface Evolver Bench74.4%—
Epoch Capabilities Index153.92—

Math Grok 4.5 leads

Grok 4.5: 60.9 (#35), Qwen3-4B: 29.7 (#240)

BenchmarkGrok 4.5Qwen3-4B
OTIS Mock AIME 2024-202597.8%52.2%
FrontierMath (Tiers 1-3)57.2%—
FrontierMath Tier 424.4%—
MathArena Final-Answer Competitions—38.5%
ProofBench31%—
LMArena Math1459—

Knowledge Grok 4.5 leads

Grok 4.5: 62.3 (#24), Qwen3-4B: 33.0 (#208)

BenchmarkGrok 4.5Qwen3-4B
GPQA Diamond93.4%52.3%
SimpleQA Verified48.3%—
Vectara Hallucination Rate—5.7%
LMArena Expert1466—

Multimodal Not comparable

Grok 4.5: 37.6 (#72), Qwen3-4B: —

BenchmarkGrok 4.5Qwen3-4B
LMArena Vision1288—
Blueprint-Bench 227.3%—
Furniture Assembly22.5%—
LMArena Document1452—

Multilingual Not comparable

Grok 4.5: 54.4 (#42), Qwen3-4B: —

BenchmarkGrok 4.5Qwen3-4B
LMArena Non-English1440—
LMArena Chinese1496—
LMArena French1456—
LMArena German1446—
LMArena Japanese1428—
LMArena Korean1404—
LMArena Russian1448—
LMArena Spanish1450—

Instruction Following Not comparable

Grok 4.5: 76.0 (#48), Qwen3-4B: —

BenchmarkGrok 4.5Qwen3-4B
LMArena Instruction Following1446—

Long Context Not comparable

Grok 4.5: 44.8 (#56), Qwen3-4B: —

BenchmarkGrok 4.5Qwen3-4B
LMArena Longer Query1463—

Writing & Preference Not comparable

Grok 4.5: 65.8 (#42), Qwen3-4B: —

BenchmarkGrok 4.5Qwen3-4B
LMArena Text1448—
LMArena Creative Writing1442—
EQ-Bench Creative Writing1579—
LMArena Multi-Turn1456—

Frequently asked questions

Is Grok 4.5 better than Qwen3-4B?

Grok 4.5 is the stronger model overall, scoring 55.0 to 31.9 on the Noometry Index.

How many benchmarks do Grok 4.5 and Qwen3-4B share?

3 benchmarks have published results for both models. Grok 4.5 has 52 scored results on Noometry and Qwen3-4B has 6.

Related comparisons

Go deeper