Grok 4.6 vs Grok 4.7
Different efforts: medium · low · 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 4.642.1
Grok 4.739.2
Arena Text · overall Arena Text · overall: Human pairwise preferences in the stated text vote pool
Grok 4.6 · high1,454
Grok 4.7 · xhigh1,444
Arena Text · Chinese Arena Text · chinese: Human pairwise preferences in the stated text vote pool
Grok 4.6 · high1,475.1
Grok 4.7 · xhigh1,452.7
Arena Text · English Arena Text · english: Human pairwise preferences in the stated text vote pool
Grok 4.6 · high1,432.5
Grok 4.7 · xhigh1,412.3
Arena Text · French Arena Text · french: Human pairwise preferences in the stated text vote pool
Grok 4.6 · high1,456.9
Grok 4.7 · xhigh1,442.8
Arena Text · German Arena Text · german: Human pairwise preferences in the stated text vote pool
Grok 4.6 · high1,431.9
Grok 4.7 · xhigh1,353.3
Arena Text · instruction following Arena Text · instruction following: Human pairwise preferences in the stated text vote pool
Grok 4.6 · high1,430.6
Grok 4.7 · xhigh1,407.6
Arena Text · Japanese Arena Text · japanese: Human pairwise preferences in the stated text vote pool
Grok 4.6 · high1,380.5
Grok 4.7 · xhigh1,355.1
Arena Text · Korean Arena Text · korean: Human pairwise preferences in the stated text vote pool
Grok 4.6 · high1,395.8
Grok 4.7 · xhigh1,385.1
Arena Text · non-English Arena Text · non english: Human pairwise preferences in the stated text vote pool
Grok 4.6 · high1,418.9
Grok 4.7 · xhigh1,386.2
Arena Text · Polish Arena Text · polish: Human pairwise preferences in the stated text vote pool
Grok 4.6 · high1,429.5
Grok 4.7 · xhigh1,411.3
Arena Text · Russian Arena Text · russian: Human pairwise preferences in the stated text vote pool
Grok 4.6 · high1,421.8
Grok 4.7 · xhigh1,399.9
Arena Text · Spanish Arena Text · spanish: Human pairwise preferences in the stated text vote pool
Grok 4.6 · high1,403.7
Grok 4.7 · xhigh1,404.7
AA Omniscience Index AA Omniscience Index: Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty
Grok 4.628
Grok 4.729.6
SimpleQA Verified v1 SimpleQA Verified v1: Accuracy on short factual questions with verified answers, answered without web access
Grok 4.652.9
Grok 4.759.4
CritPt (physics) CritPt (physics): Research-level physics problems that need many careful reasoning steps
Grok 4.617.7
Grok 4.713.4
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 4.642.8
Grok 4.742.2
AA Omniscience · Accuracy AA Omniscience · Accuracy: Measures correct answers on the AA Omniscience factual knowledge test
Grok 4.641.9
Grok 4.749.7
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 4.624
Grok 4.739.9
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 4.681
Grok 4.780
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 4.617.8
Grok 4.721.4
MLCR-AA MLCR-AA: Models synthesize long medical records; the headline credits concise answers judged both accurate and complete
Grok 4.6 · high12.2
Grok 4.7 · xhigh15
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 4.694.8
Grok 4.782.3
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 4.687.5
Grok 4.763.3
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 4.667.4
Grok 4.725
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 4.661.3
Grok 4.716.7
ARC-AGI-3 · semi-private ARC-AGI-3 · semi-private: Measures interactive game solving through relative human action efficiency
Grok 4.6 · xhigh2.1
Grok 4.70.2
Chess Puzzles Chess Puzzles: Models choose the best next move from a novel chess position
Grok 4.6 · high33
Grok 4.7 · xhigh41
FrontierMath Tier 4 (v2) FrontierMath Tier 4 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
Grok 4.6 · xhigh31.7
Grok 4.7 · xhigh17.1
FrontierMath Tiers 1–3 (v2) FrontierMath Tiers 1–3 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
Grok 4.6 · xhigh66
Grok 4.7 · xhigh53
Mystery Game Puzzles Mystery Game Puzzles: Models choose the best move in puzzle positions from an undisclosed game
Grok 4.6 · xhigh34
Grok 4.7 · xhigh29
Vals LegalBench Vals LegalBench: Legal reasoning and legal-language understanding on LegalBench
Grok 4.686.3
Grok 4.784.4
Vals MedCode Vals MedCode: Medical coding accuracy on patient encounters
Grok 4.644.7
Grok 4.749.6
Vals MedScribe Vals MedScribe: Accuracy of clinical notes produced from consultation transcripts
Grok 4.686.5
Grok 4.789.4
Vals Legal Research Bench Vals Legal Research Bench: Grounded legal research; a task passes only if all criteria pass
Grok 4.648.1
Grok 4.747.1
Vals Science Index Vals Science Index: Scientific problem solving across the index’s component evaluations
Grok 4.638.1
Grok 4.734.8
FrontierMath Erdos FrontierMath Erdos: Models solve expert-written advanced mathematics problems on the named tier and edition
Grok 4.6—
Grok 4.7 · xhigh0
MathArena · ArXivMath · August 2026 MathArena · arxiv · 08/2026: Answer accuracy on the named mathematics competition under the recorded effort
Grok 4.6—
Grok 4.7 · xhigh46.5
MathArena · ArXivMath · June 2026 MathArena · arxiv · 06/2026: Answer accuracy on the named mathematics competition under the recorded effort
Grok 4.6—
Grok 4.7 · xhigh70.1
MathArena · ArXivMath · May 2026 MathArena · arxiv · 05/2026: Answer accuracy on the named mathematics competition under the recorded effort
Grok 4.6—
Grok 4.7 · xhigh60.8
MathArena · ArXivMath · September 2026 MathArena · arxiv · 09/2026: Answer accuracy on the named mathematics competition under the recorded effort
Grok 4.6—
Grok 4.7 · xhigh45.3
MathArena · BrokenArXiv · August 2026 MathArena · arxiv_false · 08/2026: Answer accuracy on the named mathematics competition under the recorded effort
Grok 4.6—
Grok 4.7 · xhigh35.1
MathArena · BrokenArXiv · June 2026 MathArena · arxiv_false · 06/2026: Answer accuracy on the named mathematics competition under the recorded effort
Grok 4.6—
Grok 4.7 · xhigh52.8
MathArena · BrokenArXiv · May 2026 MathArena · arxiv_false · 05/2026: Answer accuracy on the named mathematics competition under the recorded effort
Grok 4.6—
Grok 4.7 · xhigh40
MathArena · BrokenArXiv · September 2026 MathArena · arxiv_false · 09/2026: Answer accuracy on the named mathematics competition under the recorded effort
Grok 4.6—
Grok 4.7 · xhigh27.7
| Benchmark | Grok 4.6 | Grok 4.7 |
|---|---|---|
| Humanity's Last Exam | 42.1 | 39.2 |
| Arena Text · overall | 1,454 · high | 1,444 · xhigh |
| Arena Text · Chinese | 1,475.1 · high | 1,452.7 · xhigh |
| Arena Text · English | 1,432.5 · high | 1,412.3 · xhigh |
| Arena Text · French | 1,456.9 · high | 1,442.8 · xhigh |
| Arena Text · German | 1,431.9 · high | 1,353.3 · xhigh |
| Arena Text · instruction following | 1,430.6 · high | 1,407.6 · xhigh |
| Arena Text · Japanese | 1,380.5 · high | 1,355.1 · xhigh |
| Arena Text · Korean | 1,395.8 · high | 1,385.1 · xhigh |
| Arena Text · non-English | 1,418.9 · high | 1,386.2 · xhigh |
| Arena Text · Polish | 1,429.5 · high | 1,411.3 · xhigh |
| Arena Text · Russian | 1,421.8 · high | 1,399.9 · xhigh |
| Arena Text · Spanish | 1,403.7 · high | 1,404.7 · xhigh |
| AA Omniscience Index | 28 | 29.6 |
| SimpleQA Verified v1 | 52.9 | 59.4 |
| CritPt (physics) | 17.7 | 13.4 |
| AA Intelligence Index 4.3.2 | 42.8 | 42.2 |
| AA Omniscience · Accuracy | 41.9 | 49.7 |
| AA Omniscience · Hallucination Rate | 24 | 39.9 |
| AA-LCR (long context) 1.1 | 81 | 80 |
| GDP.pdf · AA all-pass | 17.8 | 21.4 |
| MLCR-AA | 12.2 · high | 15 · xhigh |
| ARC-AGI-1 · public evaluation | 94.8 | 82.3 |
| ARC-AGI-1 · semi-private | 87.5 | 63.3 |
| ARC-AGI-2 · public evaluation | 67.4 | 25 |
| ARC-AGI-2 · semi-private | 61.3 | 16.7 |
| ARC-AGI-3 · semi-private | 2.1 · xhigh | 0.2 |
| Chess Puzzles | 33 · high | 41 · xhigh |
| FrontierMath Tier 4 (v2) | 31.7 · xhigh | 17.1 · xhigh |
| FrontierMath Tiers 1–3 (v2) | 66 · xhigh | 53 · xhigh |
| Mystery Game Puzzles | 34 · xhigh | 29 · xhigh |
| Vals LegalBench | 86.3 | 84.4 |
| Vals MedCode | 44.7 | 49.6 |
| Vals MedScribe | 86.5 | 89.4 |
| Vals Legal Research Bench | 48.1 | 47.1 |
| Vals Science Index | 38.1 | 34.8 |
| FrontierMath Erdos | — | 0 · xhigh |
| MathArena · ArXivMath · August 2026 | — | 46.5 · xhigh |
| MathArena · ArXivMath · June 2026 | — | 70.1 · xhigh |
| MathArena · ArXivMath · May 2026 | — | 60.8 · xhigh |
| MathArena · ArXivMath · September 2026 | — | 45.3 · xhigh |
| MathArena · BrokenArXiv · August 2026 | — | 35.1 · xhigh |
| MathArena · BrokenArXiv · June 2026 | — | 52.8 · xhigh |
| MathArena · BrokenArXiv · May 2026 | — | 40 · xhigh |
| MathArena · BrokenArXiv · September 2026 | — | 27.7 · xhigh |
Specs
| Spec | Grok 4.6 | Grok 4.7 |
|---|---|---|
| Input, per 1M | $2 | $2 |
| Output, per 1M | $6 | $6 |
| Cost per task? | $1.14 | $1.25 |
| Speed, tok/s | 68 | 80 |
| Context | 500K | 500K |
| Weights | Closed | Closed |
| Released | Aug 2026 | Sep 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 4.6 → Grok 4.7
Reasoning cannot be disabled›
Reasoning cannot be disabled; choose low, medium, high or xhigh effort.
Penalties and stop return error›
presencePenalty, frequencyPenalty and stop cannot be used with reasoning models; requests containing them return an error.
Set prompt_cache_key›
Set prompt_cache_key or the Chat Completions x-grok-conv-id header to keep a conversation on the same server and improve cache hits.
Pass reasoning items back unchanged · Responses›
Responses always returns encrypted reasoning even without include; pass the reasoning items back unchanged in the next request.
10% premium · US regional endpoint›
The US regional endpoint keeps inference in the United States and adds a 10% token-price premium.
Grok 4.7 → Grok 4.6
Nothing known.