Grok 4.5 Pricing

Grok 4.5 API pricing

Grok 4.5 API pricing is $2 per million input tokens and $6 per million output tokens, verified November 1, 2026.

Input (per 1M tokens)

$2

Output (per 1M tokens)

$6

Cached input (per 1M tokens)

$0.3

Context window

500,000 tokens

Released

July 8, 2026

Capabilities

Grok 4.5 supports extended reasoning, tool use.

No Grok 4.5 price changes have been recorded yet.

No price snapshots recorded yet.

Frequently asked questions

Short answers to the most common questions about Grok 4.5 pricing, limits, and capabilities.

Notes

2026-09-22 CACHE READ CORRECTED $0.50 -> $0.30 per MTok. Input $2.00 / Output $6.00 unchanged. This field was flagged six consecutive weekly runs (Aug 17 through Sep 21) with the evidence pointing consistently one way. Resolving it now rather than letting a seventh run pass. EVIDENCE: multiple independent named aggregators (mem0.ai, benchlm.ai, spheron.network, sentisight.ai, layer3labs, costgoat) report $0.30/M for Grok 4.5 cached input. By the Sep 21 run not one source argued for $0.50. A mechanism for the original error was identified on 2026-09-07: the $0.50 figure is Grok 4.6's cached rate being mis-attributed to 4.5 — several sources state the contrast explicitly ('Grok 4.6 cached input is $0.50/M, while Grok 4.5 cached input is $0.30/M'). This clears the two-independent-named-sources bar. No primary docs.x.ai read was obtainable (egress allowlist blocks direct fetches, and the browser pane requires site approval). TIERED PRICING CAVEAT: Grok 4.5 tiers at a 200K prompt threshold. $0.30 is the cached rate at or below 200K; above 200K it is $0.60. This row stores the sub-200K bracket, consistent with Input $2.00 / Output $6.00 also being the sub-200K figures. IMPACT: the calculator was overstating Grok 4.5 cache reads by 67%. Now corrected. If a primary docs.x.ai read later shows $0.50, revert. [2026-09-28] Verification attempt produced no clear result — recheck next run. [2026-10-09] Verification attempt produced no clear result — recheck next run. Input $2 / Output $6 consistent across sources. Cache read conflict resurfaced: pricepertoken, benchlm and apidog snippets now report $0.50 cached (benchlm attributes it to docs.x.ai), with a >200K cached rate of $1.00 (vs $0.60 in the 09-22 note). Cache Read LEFT at $0.30 — not reverting without a primary docs.x.ai read. [2026-10-10] Verification attempt produced no clear result — recheck next run. Input $2 / Output $6 confirmed again. Cache read: most guides this run (incl. one citing xAI's launch post / Grok 4.5 docs) now say $0.50; only trevorfox (verified 2026-07-24) still shows $0.30. Second straight run where the majority favours $0.50. Cache Read LEFT at $0.30 per no-primary-read rule. ACTION FOR KHALED: check docs.x.ai — likely revert to $0.50. [2026-10-11] Verification attempt produced no clear result — recheck next run. Input $2 / Output $6 confirmed (benchlm, eesel, cognee). Cache read still split: one source $0.30, another $0.50 ('75% discount off standard'). Cache Read LEFT at $0.30. Third run running with this conflict — manual docs.x.ai check needed. [2026-10-13] Verification attempt produced no clear result — recheck next run. Input $2 / Output $6 confirmed. Cache read: CometAPI and BenchLM (BenchLM citing the Grok 4.5 docs) say $0.50; nOps says $0.30. Fourth straight run with the majority at $0.50. Cache Read LEFT at $0.30 — manual docs.x.ai check needed. [2026-10-15] Verification attempt produced no clear result — recheck next run. Input $2 / Output $6 confirmed (benchlm, codersera, layer3labs). Cache read still split: layer3labs $0.30 (85% off), others $0.50 (75% off). Cache Read LEFT at $0.30 — fifth straight run; manual docs.x.ai check needed. [2026-10-19] Verification attempt produced no clear result — recheck next run. Input $2 / Output $6 confirmed again across aggregators. Cache read still split: $0.30 (layer3labs plus two others) vs $0.50 (one source). Cache Read LEFT at $0.30 — sixth straight run; manual docs.x.ai check needed. [2026-10-27] Verification attempt produced no clear result — recheck next run. Input $2 / Output $6 consistent across aggregators (morphllm, cloudzero, cometapi). Cache read still split $0.30 vs $0.50 across two tables. Cache Read LEFT at $0.30 — manual docs.x.ai check needed. [2026-10-30] Verification attempt produced no clear result — recheck next run. Input $2 / Output $6 confirmed (cloudzero, techjacksolutions, layer3labs). Cache read still split: layer3labs/spheron/sentisight-type trackers $0.50 vs techjacksolutions/Grokified $0.30 (Grokified: '$0.30 for 4.5 vs $0.50 for 4.6/4.7'). One source says xAI never published a 4.5 cached rate. Cache Read LEFT at $0.30 — manual docs.x.ai check needed. [2026-11-01] Verification attempt produced no clear result — recheck next run. Input $2 / Output $6 / 500K context confirmed (cometapi, benchlm). Cache read: this run's snippets (cometapi, benchlm) give $0.50; no $0.30 source surfaced. Still no docs.x.ai read. Cache Read LEFT at $0.30 — manual docs.x.ai check needed.