Grok 4.20 (reasoning) vs Grok 4.20 (non-reasoning)
Facts only. This page does not pick a winner. For a picked winner, see the use case comparison on Tools.
| Grok 4.20 (reasoning) | Grok 4.20 (non-reasoning) | |
|---|---|---|
| API model IDs | grok-4.20-0309-reasoning, grok-4.20, grok-4.20-reasoning, grok-4.20-reasoning-latest | grok-4.20-0309-non-reasoning, grok-4.20-non-reasoning, grok-4.20-non-reasoning-latest |
| Context window | 1,000,000 | 1,000,000 |
| Lifecycle | active | active |
| License | proprietary | proprietary |
| Trains on API data | No | No |
| Zero data retention | yes | yes |
| Input per 1M tokens | $1.25 (verified 2026-09-08) | $1.25 (verified 2026-09-08) |
Frequently asked questions
- Which has the larger context window, Grok 4.20 (reasoning) or Grok 4.20 (non-reasoning)?
- Grok 4.20 (reasoning) holds 1,000,000 tokens and Grok 4.20 (non-reasoning) holds 1,000,000. Both hold 1,000,000 tokens. The context window counts the prompt, any attached files, and the reply together.
- Which is cheaper, Grok 4.20 (reasoning) or Grok 4.20 (non-reasoning)?
- Per million input tokens, Grok 4.20 (reasoning) was $1.25 (verified 2026-09-08) and Grok 4.20 (non-reasoning) was $1.25 (verified 2026-09-08). Input pricing is the same on both. Output tokens cost $2.5 and $2.5 respectively, and output volume usually decides the bill.
- Does Grok 4.20 (reasoning) or Grok 4.20 (non-reasoning) train on API data?
- Grok 4.20 (reasoning) does not train on API data by default, and Grok 4.20 (non-reasoning) does not train on API data by default. Zero data retention is yes for Grok 4.20 (reasoning) and yes for Grok 4.20 (non-reasoning). Verified 2026-09-08 and 2026-09-08 against the providers' own documentation.
- Is Grok 4.20 (reasoning) or Grok 4.20 (non-reasoning) being retired?
- Grok 4.20 (reasoning) is active. Grok 4.20 (non-reasoning) is active. Lifecycle checked 2026-09-08 and 2026-09-08. Pin a dated snapshot ID rather than an alias if a retirement would break you.