Grok 4
Last updated: April 22, 2026
Specs
| Vendor | xAI |
|---|
| Released | July 9, 2025 |
|---|
| Context window | 256K tokens |
|---|
| Input | $3 / M tokens |
|---|
| Output | $15 / M tokens |
|---|
Benchmarks
| mmlu | 87.5 |
|---|
| gpqa | 57 |
|---|
| humanEval | 85.1 |
|---|
| arenaElo | 1341 |
|---|
| sweBench | 50 |
|---|
| sweBenchPro | 40 |
|---|
| liveCodeBench | 66 |
|---|
| aiderPolyglot | 62 |
|---|
| terminalBench | 35 |
|---|
| mmluPro | 80 |
|---|
| hle | 14 |
|---|
| aime | 88 |
|---|
| math500 | 94 |
|---|
| mmmu | 72 |
|---|
Strengths
- Realtime access to X (Twitter) data
- Less-restrictive tone for edgy use cases
- Good general reasoning
Weaknesses
- Smaller developer community
- Fewer third-party integrations
- Geopolitical concerns for some enterprises
Best for
- Social listening and X-based agents
- News and trends summarization
- Content moderation at the edge
Head-to-head comparisons
Compare with other AI models