Model comparison
Grok 4.6 vs Grok 4.7
Two SpaceXAI tiers compared on the figures that decide which one a workload actually needs.
Grok 4.6: profile checked October 25, 2026Grok 4.7: profile checked October 25, 2026Individual provider fields may changeEvidence confidence: Medium — see receipts below
SpaceXAI
Grok 4.6
Frontier
SpaceXAI
Grok 4.7
Frontier
| Specification | Grok 4.6 | Grok 4.7 |
|---|---|---|
| Provider | ||
| Provider | SpaceXAI | SpaceXAI |
| Tier | ||
| Tier | Frontier | Frontier |
| Context window | ||
| Context window | 500K | 500K |
| Max output | ||
| Max output | Not verifiedUnverified | Not verifiedUnverified |
| Input / 1M tokens | ||
| Input / 1M tokens | $2 | $2 |
| Output / 1M tokens | ||
| Output / 1M tokens | $6 | $6 |
| Weights | ||
| Weights | Closed | Closed |
| Parameters | ||
| Parameters | Not verifiedUnverified | Not verifiedUnverified |
| Reasoning levels | ||
| Reasoning levels | low, medium, high, xhigh | low, medium, high, xhigh |
| Modalities | ||
| Modalities | text, image | text, image |
| API model id | ||
| API model id | grok-4.6 | grok-4.7 |
| Released | ||
| Released | August 12, 2026 | September 21, 2026 |
| Artificial Analysis Intelligence Index [xhigh] · source date Grok 4.7: 2026-10-01 | ||
| Artificial Analysis Intelligence Index [xhigh] · source date Grok 4.7: 2026-10-01 | Not verifiedUnverified | 46 |
| DeepSWE 1.1 [high] · source date Grok 4.7: 2026-09-21 | ||
| DeepSWE 1.1 [high] · source date Grok 4.7: 2026-09-21 | Not verifiedUnverified | 71 |
| Terminal-Bench 4.0 [xhigh] · source date Grok 4.7: 2026-09-26 | ||
| Terminal-Bench 4.0 [xhigh] · source date Grok 4.7: 2026-09-26 | Not verifiedUnverified | 25.8 |
Dates beside benchmark scores refer to the linked source’s publication or capture date unless that source explicitly states when it measured the result. A shared URL alone does not establish comparable runs; unverified setups are marked. Prices are USD per million tokens at standard rates, excluding batch and caching discounts. Values we could not confirm from the provider are shown as “Not verified” rather than estimated.
Benchmark receipts
Where each score comes from, and how far it can be compared across models.
- Source date2026-10-01Artificial Analysis (Grok 4.7 (xhigh))
Composite index blending reasoning, knowledge, and coding evals into one 0–100 score.
Comparable with caveatOnly same-version scores are comparable. v4.3.2 was rebaselined (the top score fell from 66 on v4.1.1 to ~58), so v4.1.1 figures must not be compared with v4.3.2 figures — check the cited source and its date.
- Source date2026-09-26Artificial Analysis Terminal-Bench 4.0 (independent AA run, part of Intelligence Index v4.3.2)
Agentic terminal work: long multi-step tasks executed in a sandboxed shell environment.
Comparable with caveatDo not directly compare scores across Terminal-Bench versions: 4.0 revises and prunes the 3.0 task set and changes environment resources, requiring fresh trials. Within 4.0, harness, time limit, tools, sampling and resource settings must match; maintainer-agent and Artificial Analysis runs are not interchangeable.
- Source date2026-09-21xAI — Grok 4.7 announcement (vendor-reported; separate from the Datacurve publisher capture)
Agentic software engineering: pass@1 on real repo tasks with reported cost per completed task.
Directly comparable
Pricing tiers
Grok 4.6: Standard USD/MTok input/output $2/$6 below 200K prompt tokens; at 200K or more the full request bills $4/$12. Cached input $0.50/$1.00 for short/long requests. Batch discounts all token types by 20%; available regions and team access still apply. API Priority is 2x Standard for all token types and is billed only when the response confirms service_tier priority; it does not combine with Batch. US regional inference adds 10% to input/output/cache rates. Server tools have separate fees: web search/code execution/attachments $5 per 1,000 calls; X search $5 per 1,000 fetched posts or $10 per 1,000 profiles; collections search $2.50 per 1,000 calls. Search can fetch multiple billed items; reasoning and media tokens are also billable.
Grok 4.7: Standard USD/MTok input/output $2/$6 below 200K prompt tokens; at 200K or more the full request bills $4/$12. Cached input $0.50/$1.00 for short/long requests. Batch discounts all token types by 20%; available regions and team access still apply. API Priority is 2x Standard for all token types and is billed only when the response confirms service_tier priority; it does not combine with Batch. US regional inference adds 10% to input/output/cache rates. Cursor/Grok Build Fast is a separate plan-billed option: $4/$12 with $1 cached input below 200K, $6/$18 with $1.50 cached input above 200K. Its exact 200K boundary is ambiguous in the Fast table; confirm that edge. It is unavailable on the public xAI API or Grok Build free tier. Server tools have separate fees: web search/code execution/attachments $5 per 1,000 calls; X search $5 per 1,000 fetched posts or $10 per 1,000 profiles; collections search $2.50 per 1,000 calls. Search can fetch multiple billed items; reasoning and media tokens are also billable.
Grok 4.6
Grok 4.6 was xAI's August 2026 coding flagship, superseded by Grok 4.7 (2026-09-21) at the same $2/$6 price.
Best for
- Coding agents
- Chat and knowledge
- Cost-sensitive frontier work
Watch out
Reasoning defaults high and cannot be disabled; presencePenalty, frequencyPenalty and stop error. xhigh is supported. Batch is now listed with a 20% token discount. API Priority and the Grok 4.7 Fast partner route have different eligibility and billing. Current publisher documentation uses SpaceXSI branding for the xAI service. A separate numeric text-output ceiling remains unverified. Still listed as an earlier model; current selection docs recommend Grok 4.7. Historical benchmark identities and dates remain separate.
Grok 4.7
Grok 4.7 is xAI's flagship since 2026-09-21 — a larger new base model with longer RL on multi-hour tasks, at Grok 4.6's $2/$6 price.
Best for
- Coding agents
- Cost-sensitive frontier work
- Cursor and GitHub Copilot users
Watch out
Reasoning defaults high and cannot be disabled; presencePenalty, frequencyPenalty and stop error. xhigh is supported. Batch is now listed with a 20% token discount. API Priority and the Grok 4.7 Fast partner route have different eligibility and billing. Current publisher documentation uses SpaceXSI branding for the xAI service. The publisher explicitly states no separate text-output limit; the 500K context still constrains requests. Preserve encrypted reasoning and server-tool outputs unchanged across Responses turns. Copilot access depends on its plan and rollout; historical benchmark results are not remeasured here.
When the cheaper one wins
Both charge $6 per million output tokens at standard rates, so output price is not the split. Prefer the row whose “best for” line matches the job, then confirm the live vendor rate.
Run the model pickerEvidence confidence: Medium
How strong and complete the evidence behind this comparison is — not a prediction of which model is better.
- Pricing verified on both sides — Input and output rates are verified for both models.
- 3/5 core specs verified on both sides — Not published for at least one side: max output, parameter count.
- No directly aligned benchmark setup — No benchmark has matching cited sources, source dates, units, effort and recorded run details on both models.
- Verified within the last 90 days — Older of the two catalog checks was 24 days ago.
- Both models carry source citations — Each side has at least two catalog sources on record.
Source receipts
Each catalog figure was checked against the provider or an independent second source on the date shown.
- Grok 4.6: Whole-record check: current publisher specifications, licence, access or tariff (accessed 2026-10-25)
- Grok 4.6: Whole-record check: current publisher specifications, licence, access or tariff (accessed 2026-10-25)
- Grok 4.6: Whole-record check: current publisher specifications, licence, access or tariff (accessed 2026-10-25)
- Grok 4.6: xAI — Grok 4.6 (accessed 2026-08-29)
- Grok 4.6: xAI pricing (grok-4.6 $2/$6, 500K) (accessed 2026-08-29)
- Grok 4.7: Whole-record check: current publisher specifications, licence, access or tariff (accessed 2026-10-25)
- Grok 4.7: Whole-record check: current publisher specifications, licence, access or tariff (accessed 2026-10-25)
- Grok 4.7: Whole-record check: current publisher specifications, licence, access or tariff (accessed 2026-10-25)
- Grok 4.7: Whole-record check: current publisher specifications, licence, access or tariff (accessed 2026-10-25)
- Grok 4.7: Whole-record check: current publisher specifications, licence, access or tariff (accessed 2026-10-25)
- Grok 4.7: Whole-record check: current publisher specifications, licence, access or tariff (accessed 2026-10-25)
- Grok 4.7: xAI — Grok 4.7 (accessed 2026-09-26)
Related comparisons
Common questions
Grok 4.6 vs Grok 4.7
Answered from the verified figures on this page rather than general guidance.
Is Grok 4.6 or Grok 4.7 cheaper for input?
Is Grok 4.6 or Grok 4.7 cheaper for output?
Which has the larger context window, Grok 4.6 or Grok 4.7?
Do Grok 4.6 and Grok 4.7 support the same reasoning levels?
Should I use Grok 4.6 or Grok 4.7?
Next step
Choosing between them
The verified figures that separate this pair, computed from the catalog rather than restated boilerplate.
These two models match on every verified catalog figure — the decision comes down to provider preference, availability, and workload fit rather than published numbers.
Choosing a harness rather than a model — Cursor, Copilot, Claude Code, Windsurf, or Lovable? Compare agentic harnesses · Latest releases.