Grok 4.7 is SpaceXAI's flagship, released Sep 21, 2026 on a larger base model; Grok 4.3 is the reasoning workhorse it sits above. The older model is cheaper on every rate and, unusually, has twice the context window.
Grok 4.3 is cheaper on input, output and cache hits ($1.25/$2.50 vs $2.00/$6.00 per 1M) and accepts 1M tokens against Grok 4.7's 500K. Pay for 4.7 only when its larger base model wins on your evals.
| Grok 4.7 | Grok 4.3 | |
|---|---|---|
| Provider | SpaceXAI | SpaceXAI |
| Tier | Flagship | Mid-tier |
| Input / 1M tokens | $2.00 | $1.25 |
| Output / 1M tokens | $6.00 | $2.50 |
| Cache hit / 1M | $0.50 | $0.20 |
| Cache write / 1M | — | — |
| Context window | 500K | 1M |
| Notes | New flagship (Sep 21, 2026); larger base model, replaces 4.6 at the same price; ≥200K tokens: $4/$12, cache $1.00 | Reasoning workhorse; replaced Grok 4; ≥200K tokens: $2.50/$5, cache $0.40 |
Grok 4.3 is 38% cheaper on input tokens ($1.25 vs $2.00 per 1M). Grok 4.3 is 2.4× cheaper on output tokens ($2.50 vs $6.00 per 1M).
| Workload | Grok 4.7 | Grok 4.3 | Difference |
|---|---|---|---|
| Support chatbot 10,000 conversations × (2,000 input + 400 output tokens) | $64.00 | $35.00 | Grok 4.3 saves $29.00 |
| Document Q&A 1,000 questions × (30,000 input + 800 output tokens) | $64.80 | $39.50 | Grok 4.3 saves $25.30 |
| Coding agent session 4M input tokens (90% cache hits) + 150,000 output tokens | $3.50 | $1.59 | Grok 4.3 saves $1.91 |
Standard rates per 1M tokens from data on this site. The coding-agent row assumes 90% of input is read from the prompt cache (full input price where a model has no cache rate). Cache-write fees, batch discounts and long-context surcharges listed in the notes are not included. Updated November 2, 2026.
you need SpaceXAI's strongest model and your evals show it beating 4.3 on hard tasks, or you were on Grok 4.6 — 4.7 replaced it at the same price.
you want lower cost per token, cheaper cache hits, or prompts beyond 500K tokens — 4.3 is the only one of the two that goes up to 1M.
Both models bill prompts of 200K tokens or more at a higher rate, and xAI's model page says it applies to every token in that request: Grok 4.7 rises to $4/$12, Grok 4.3 to $2.50/$5. Grok 4.7's 500K window means only 200K–500K prompts reach that tier; on Grok 4.3, prompts from 200K up to 1M do.
Grok 4.7's cache hit is $0.50 per 1M — a quarter of its input rate — versus $0.20 on 4.3. For agents resending long contexts, 4.7's cached input costs 2.5× more, on top of the 2.4× output gap shown in the cost table.
If you were on Grok 4.6, moving to 4.7 is free on price. If you are on 4.3, upgrading raises the bill by roughly 1.6× to 2.4× depending on output share, so reserve 4.7 for requests that need it. Grok 4.3 itself replaced Grok 4.
Yes. Grok 4.7 costs $2.00 input and $6.00 output per 1M; Grok 4.3 costs $1.25 and $2.50.
Grok 4.3, with 1M tokens, versus 500K on Grok 4.7.
Yes. From 200K prompt tokens, Grok 4.7 bills $4/$12 and Grok 4.3 $2.50/$5 per 1M for the whole request.
How we choose: prices come from each vendor's official pricing page and are re-checked every week. We take no affiliate fees, so rankings aren't for sale.