Best cell per row highlighted. Null means undisclosed — never counted as zero.
| Provider | xAI | xAI |
| Family / tier | Grok Build · Coder | Grok 4 · Flagship |
| Status | GA | GA |
| Released | 2026-05-19 | 2026-04-29 |
| Knowledge cutoff | 2024-11 | 2025-12 |
| Context window | 256K | 1M |
| Max output tokens | 256K | — |
| Input modalities | text, image | text, image, video |
| Reasoning mode | always | always |
| Open weights | No | No |
| Input | $1 | $1.25 |
| Output | $2 | $2.5 |
| Cached input | $0.2 | $0.2 |
| Batch input | — | — |
| Free tier | No | Yes |
| Output speed (tok/s) | — | 181.5 |
| Time to first token (s) | — | 19.69 |
| Latency tier | slow | slow |
| Trains on inputs | No | Yes |
| License | Proprietary | Proprietary |
| Clouds | — | Azure AI Foundry |
| Compliance | — | — |
| Unified score | 6.7 | 7.7 |
| Decision Maker | 7 | 7.5 |
| Domain Strategist | 7 | 8 |
| Finance Lead | 7.5 | 8.5 |
| Domain Practitioner | 7 | 7.5 |
| Power User | 6.5 | 7.5 |
| Skeptic | 6 | 6.5 |
| Value score | 7.5 | 8.5 |
| GPQA Diamond | — | 90 |
| SWE-bench Verified | 70.8 | — |
| IFEval | — | 81 |
| TAU-bench | — | 98 |
| Artificial Analysis Index | — | 53 |
Scores link to their sources. Missing cells mean the vendor hasn't disclosed a result — honesty over padding.