Grok 4.5 vs Grok 4.6
Different efforts: high · medium · Compare all at high
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
Grok 4.542.7
Grok 4.642.1
Arena Text · overall Arena Text · overall: Human pairwise preferences in the stated text vote pool
Grok 4.51,466
Grok 4.6 · high1,454
Arena Text · Chinese Arena Text · chinese: Human pairwise preferences in the stated text vote pool
Grok 4.51,494.6
Grok 4.6 · high1,475.1
Arena Text · English Arena Text · english: Human pairwise preferences in the stated text vote pool
Grok 4.51,453
Grok 4.6 · high1,432.5
Arena Text · French Arena Text · french: Human pairwise preferences in the stated text vote pool
Grok 4.51,456.6
Grok 4.6 · high1,456.9
Arena Text · German Arena Text · german: Human pairwise preferences in the stated text vote pool
Grok 4.51,448.2
Grok 4.6 · high1,431.9
Arena Text · instruction following Arena Text · instruction following: Human pairwise preferences in the stated text vote pool
Grok 4.51,445.6
Grok 4.6 · high1,430.6
Arena Text · Japanese Arena Text · japanese: Human pairwise preferences in the stated text vote pool
Grok 4.51,430.7
Grok 4.6 · high1,380.5
Arena Text · Korean Arena Text · korean: Human pairwise preferences in the stated text vote pool
Grok 4.51,405.8
Grok 4.6 · high1,395.8
Arena Text · non-English Arena Text · non english: Human pairwise preferences in the stated text vote pool
Grok 4.51,439.8
Grok 4.6 · high1,418.9
Arena Text · Polish Arena Text · polish: Human pairwise preferences in the stated text vote pool
Grok 4.51,432.5
Grok 4.6 · high1,429.5
Arena Text · Russian Arena Text · russian: Human pairwise preferences in the stated text vote pool
Grok 4.51,449.1
Grok 4.6 · high1,421.8
Arena Text · Spanish Arena Text · spanish: Human pairwise preferences in the stated text vote pool
Grok 4.51,451.3
Grok 4.6 · high1,403.7
AA Omniscience Index AA Omniscience Index: Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty
Grok 4.525.3
Grok 4.628
SimpleQA Verified v1 SimpleQA Verified v1: Accuracy on short factual questions with verified answers, answered without web access
Grok 4.548.3
Grok 4.652.9
CritPt (physics) CritPt (physics): Research-level physics problems that need many careful reasoning steps
Grok 4.515.4
Grok 4.617.7
AA Intelligence Index 4.3.2 AA Intelligence Index 4.3.2: Artificial Analysis combines tests of knowledge, reasoning, coding and tool use into one score
Grok 4.538.8
Grok 4.642.8
AA Omniscience · Accuracy AA Omniscience · Accuracy: Measures correct answers on the AA Omniscience factual knowledge test
Grok 4.551.6
Grok 4.641.9
AA Omniscience · Hallucination Rate AA Omniscience · Hallucination Rate: Measures how often attempted answers are incorrect on the AA Omniscience factual knowledge test · lower is better
Grok 4.554.1
Grok 4.624
AA-LCR (long context) 1.1 AA-LCR (long context) 1.1: Questions that need reading and connecting facts across very long documents (around 100K tokens)
Grok 4.579.3
Grok 4.681
GDP.pdf · AA all-pass GDP.pdf · AA all-pass: Models reason over professional PDF documents; an attempt passes only when every atomic criterion passes
Grok 4.518.8
Grok 4.617.8
MLCR-AA MLCR-AA: Models synthesize long medical records; the headline credits concise answers judged both accurate and complete
Grok 4.515
Grok 4.6 · high12.2
ARC-AGI-1 · public evaluation ARC-AGI-1 · public-eval: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Grok 4.594.8
Grok 4.694.8
ARC-AGI-1 · semi-private ARC-AGI-1 · semi-private: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Grok 4.585.7
Grok 4.687.5
ARC-AGI-2 · public evaluation ARC-AGI-2 · public-eval: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Grok 4.558.2
Grok 4.667.4
ARC-AGI-2 · semi-private ARC-AGI-2 · semi-private: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Grok 4.552.6
Grok 4.661.3
ARC-AGI-3 · semi-private ARC-AGI-3 · semi-private: Measures interactive game solving through relative human action efficiency
Grok 4.50.3
Grok 4.6 · xhigh2.1
Chess Puzzles Chess Puzzles: Models choose the best next move from a novel chess position
Grok 4.535
Grok 4.6 · high33
FrontierMath Tier 4 (v2) FrontierMath Tier 4 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
Grok 4.524.4
Grok 4.6 · xhigh31.7
FrontierMath Tiers 1–3 (v2) FrontierMath Tiers 1–3 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
Grok 4.557.2
Grok 4.6 · xhigh66
Vals LegalBench Vals LegalBench: Legal reasoning and legal-language understanding on LegalBench
Grok 4.586
Grok 4.686.3
Vals MedCode Vals MedCode: Medical coding accuracy on patient encounters
Grok 4.543.3
Grok 4.644.7
Vals MedScribe Vals MedScribe: Accuracy of clinical notes produced from consultation transcripts
Grok 4.586.9
Grok 4.686.5
Vals Legal Research Bench Vals Legal Research Bench: Grounded legal research; a task passes only if all criteria pass
Grok 4.538
Grok 4.648.1
Mystery Game Puzzles Mystery Game Puzzles: Models choose the best move in puzzle positions from an undisclosed game
Grok 4.5—
Grok 4.6 · xhigh34
MathArena · ArXivMath · June 2026 MathArena · arxiv · 06/2026: Answer accuracy on the named mathematics competition under the recorded effort
Grok 4.566.7
Grok 4.6—
MathArena · BrokenArXiv · June 2026 MathArena · arxiv_false · 06/2026: Answer accuracy on the named mathematics competition under the recorded effort
Grok 4.529.3
Grok 4.6—
Vals Science Index Vals Science Index: Scientific problem solving across the index’s component evaluations
Grok 4.5—
Grok 4.638.1
| Benchmark | Grok 4.5 | Grok 4.6 |
|---|---|---|
| Humanity's Last Exam | 42.7 | 42.1 |
| Arena Text · overall | 1,466 | 1,454 · high |
| Arena Text · Chinese | 1,494.6 | 1,475.1 · high |
| Arena Text · English | 1,453 | 1,432.5 · high |
| Arena Text · French | 1,456.6 | 1,456.9 · high |
| Arena Text · German | 1,448.2 | 1,431.9 · high |
| Arena Text · instruction following | 1,445.6 | 1,430.6 · high |
| Arena Text · Japanese | 1,430.7 | 1,380.5 · high |
| Arena Text · Korean | 1,405.8 | 1,395.8 · high |
| Arena Text · non-English | 1,439.8 | 1,418.9 · high |
| Arena Text · Polish | 1,432.5 | 1,429.5 · high |
| Arena Text · Russian | 1,449.1 | 1,421.8 · high |
| Arena Text · Spanish | 1,451.3 | 1,403.7 · high |
| AA Omniscience Index | 25.3 | 28 |
| SimpleQA Verified v1 | 48.3 | 52.9 |
| CritPt (physics) | 15.4 | 17.7 |
| AA Intelligence Index 4.3.2 | 38.8 | 42.8 |
| AA Omniscience · Accuracy | 51.6 | 41.9 |
| AA Omniscience · Hallucination Rate | 54.1 | 24 |
| AA-LCR (long context) 1.1 | 79.3 | 81 |
| GDP.pdf · AA all-pass | 18.8 | 17.8 |
| MLCR-AA | 15 | 12.2 · high |
| ARC-AGI-1 · public evaluation | 94.8 | 94.8 |
| ARC-AGI-1 · semi-private | 85.7 | 87.5 |
| ARC-AGI-2 · public evaluation | 58.2 | 67.4 |
| ARC-AGI-2 · semi-private | 52.6 | 61.3 |
| ARC-AGI-3 · semi-private | 0.3 | 2.1 · xhigh |
| Chess Puzzles | 35 | 33 · high |
| FrontierMath Tier 4 (v2) | 24.4 | 31.7 · xhigh |
| FrontierMath Tiers 1–3 (v2) | 57.2 | 66 · xhigh |
| Vals LegalBench | 86 | 86.3 |
| Vals MedCode | 43.3 | 44.7 |
| Vals MedScribe | 86.9 | 86.5 |
| Vals Legal Research Bench | 38 | 48.1 |
| Mystery Game Puzzles | — | 34 · xhigh |
| MathArena · ArXivMath · June 2026 | 66.7 | — |
| MathArena · BrokenArXiv · June 2026 | 29.3 | — |
| Vals Science Index | — | 38.1 |
Specs
| Spec | Grok 4.5 | Grok 4.6 |
|---|---|---|
| Input, per 1M | $2 | $2 |
| Output, per 1M | $6 | $6 |
| Cost per task? | $0.88 | $1.14 |
| Speed, tok/s | 63 | 68 |
| Context | 500K | 500K |
| Weights | Closed | Closed |
| Released | Jul 2026 | Aug 2026 |
| See images | ✓ | ✓ |
| Read PDFs | Partial | Partial |
| Hear audio | — | — |
| Watch video | — | — |
| Call tools | ✓ | ✓ |
| Force a tool call | Partial | Partial |
| Return strict JSON | ✓ | ✓ |
| Search the web | ? | ? |
| Run code | ? | ? |
| Use a computer | ? | ? |
What breaks when you switch
Grok 4.5 → Grok 4.6
Nothing known.
Grok 4.6 → Grok 4.5
Nothing known.