| Spec | Claude Opus 4 | Grok-3 |
|---|
| Provider | Anthropic | xAI |
| Release Date | May 2025 | Feb 2025 |
| Knowledge Cutoff | Mar 2025 | 2024-11-17 |
| Parameters | unknown | Undisclosed |
| Context Window | 200K | 128K |
| Max Output | — | — |
| Open Source | No | No |
| License | proprietary | proprietary |
| Tokenizer | — | — |
| Modality | — | — |
| Reasoning Model | No | No |
| Moderated | No | No |
| Price (per 1M tokens) | Claude Opus 4 | Grok-3 |
|---|---|---|
| Input | $15.00 | $3.00 |
| Output | $75.00 | $15.00 |
| Feature | Claude Opus 4 | Grok-3 |
|---|---|---|
| Text Input | ||
| Image Input (Vision) | ||
| Audio Input | ||
| Video Input | ||
| File Input | ||
| Image Output | ||
| Audio Output | ||
| Tool Use | ||
| Structured Output (JSON) | ||
| Streaming | ||
| Reasoning Tokens | ||
| Web Search | ||
| Temperature Control | ||
| Top-P Sampling | ||
| Stop Sequences | ||
| Seed (Reproducibility) | ||
| Log Probabilities |
| Benchmark | Claude Opus 4 | Grok-3 |
|---|---|---|
| IFEval | 91.5 | — |
| MMLU-Pro | 83.5 | — |
| Chatbot Arena Elo | 1365 | — |
| Benchmark | Claude Opus 4 | Grok-3 |
|---|---|---|
| HumanEval | 95.1 | — |
| SWE-bench Verified | 55.8 | — |
| Benchmark | Claude Opus 4 | Grok-3 |
|---|---|---|
| GPQA Diamond | 74.8 | 84.6 |
| BigBench-Hard | 93.2 | — |
| Benchmark | Claude Opus 4 | Grok-3 |
|---|---|---|
| MMMU | 74.2 | 78 |