Up to four models
Grok 4.5 vs Grok 4.3
On the index (GPT-4 = 100), Grok 4.5 scores 161 and Grok 4.3 135. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
Grok 4.5
161145–177
WeightsClosed
ReleasedJul 2026
Price in / out$2.0 / $6.0
Context500K
Core tests taken6 of 9
Grok 4.3
135115–156
WeightsClosed
ReleasedApr 2026
Price in / out$1.25 / $2.5
Context1M
Core tests taken1 of 9
Grok 4.5 leads Grok 4.3 by 26 points, ahead on 1 of 1 shared core tests, at 1.9× the price.
Test by testthird-party results
Grok 4.5Grok 4.3
Every result as a table
| Test | Grok 4.5 | Grok 4.3 |
|---|---|---|
| Reasoning & knowledge | ||
| LiveBench Data Analysis | 73.0% | 55.8% |
| LiveBench IF | 71.5% | 62.7% |
| LiveBench Language | 82.8% | 73.6% |
| LiveBench Reasoning | 87.2% | 70.8% |
| GPQA Diamond (Vals AI) | 92.9% | 91.4% |
| MMLU Pro | 89.2% | 85.8% |
| MMMU Pro | 61.8% | 83.1% |
| Maths | ||
| LiveBench Mathematics | 90.8% | 84.3% |
| Coding | ||
| LiveBench Agentic Coding | 56.5% | 18.5% |
| LiveBench Coding | 68.6% | 69.9% |
| Code Migration | 36.6% | 6.8% |
| LiveCodeBench | 87.4% | 84.5% |
| ProgramBench | 0.0% | 0.0% |
| SWE-bench | 86.6% | 71.4% |
| Vibe Code Bench v1.1 | 69.0% | 19.4% |
| Agents | ||
| SkillsBench | 66.0% | 40.6% |
| Professional work | ||
| Finance Agent (v2) | 48.4% | 37.7% |
| CorpFin v2 | 67.4% | 68.5% |
| EMB | 52.9% | 18.4% |
| LegalBench | 86.0% | 84.5% |
| Legal Research Bench | 38.0% | 15.4% |
| MedCode | 43.3% | 38.1% |
| MedScribe | 86.9% | 74.4% |
| MortgageTax | 61.8% | 48.3% |
| Tax Agent Bench | 24.9% | 6.2% |
| TaxEval v2 | 71.7% | 70.8% |
Tests at least two of these models have taken. Bold rows are the core tests.