Side-by-side comparison of pricing, 12 benchmarks, and generation speed.
| Metric | Llama 3.1 Instruct 405B | Grok 4.20 Beta 0309 (Non-reasoning) |
|---|---|---|
| Input ($/M tokens) | $2.75 | $2 |
| Output ($/M tokens) | $6.5 | $6 |
Data from Artificial Analysis API — 12 benchmarks
Grok 4.20 Beta 0309 (Non-reasoning) is cheaper overall. Its blended price (3:1 input/output ratio) is $3.00/M tokens vs $3.69/M for Llama 3.1 Instruct 405B.
Grok 4.20 Beta 0309 (Non-reasoning) wins 7 out of 12 benchmarks compared to 5 for Llama 3.1 Instruct 405B. See the detailed benchmark chart above for per-category results.
Grok 4.20 Beta 0309 (Non-reasoning) generates tokens faster at 189 tok/s vs 31 tok/s. However, Grok 4.20 Beta 0309 (Non-reasoning) has lower time-to-first-token (0.31s vs 0.47s).
Choose based on your priorities: Grok 4.20 Beta 0309 (Non-reasoning) for lower cost, Grok 4.20 Beta 0309 (Non-reasoning) for stronger benchmark performance, and Grok 4.20 Beta 0309 (Non-reasoning) for faster generation. For latency-sensitive apps, check the TTFT comparison above.