🚀 Released Aug 12, 2026 — $2/$6 standard, fast variant 2x

Grok 4.6

xAI (SpaceXAI)'s newest flagship, built for code, agents, and knowledge work. Scores 61 on the Artificial Analysis Intelligence Index — matching GPT-5.6 Sol — at an unchanged $2/$6 standard price.

$2.00
Input / 1M tokens
$6.00
Output / 1M tokens
500K
Context window
Text+Image
Input modalities
Last updated: | Sources: xAI Announcement · docs.x.ai Model · docs.x.ai Pricing | All Pricing · Calculator · Gemini 3.7 Flash · 日本語版

Pricing (as of Aug 12, 2026 launch)

Verified against docs.x.ai API pricing. All prices in USD per 1M tokens. Standard rates are unchanged from Grok 4.5 ($2/$6).

TierInputCached InputOutputNotes
Standard (short <200K)$2.00$0.50$6.00Default rate
Long context (≥200K)$4.00$1.00$12.00Applies to all tokens once ≥200K
Fast variant$4.00$12.00Lower latency. 2× standard
⚠️ 200K long-context boundary: Once a prompt reaches 200K tokens, the long-context rate ($4/$12) applies to all tokens in the request (per docs.x.ai). Agent loops that accumulate tool output can cross this line and repriced mid-turn — budget accordingly.
💡 Cut real costs with cached input: Cached input is $0.50 (75% off standard input). Reusing a stable prefix (system prompt, repo context) drives the effective rate down substantially. No Batch API discount for grok-4.6 (batch discounts apply only to grok-4.3/4.20 models). Tool calls are billed separately (e.g., web_search $5/1k calls, file_search $2.50/1k calls).

Primary sources: docs.x.ai/developers/pricing · docs.x.ai Grok 4.6 · x.ai/news/grok-4-6 ("Pricing starts at $2 per million input tokens and $6 per million output tokens. Additionally, there is a fast variant which is twice the price." — verified 2026-08-17)

Model Specifications

GA Release
Aug 12, 2026
Knowledge Cutoff
Feb 1, 2026
Context Window
500,000 tokens
Input Modalities
Text + Image
Output Modalities
Text
Tool Use
Function Calling ✓
Structured Outputs
Supported
Reasoning
Configurable (reasoning_effort)
Regions
us-east-1 / us-west-2
Model ID
grok-4.6

Sources: docs.x.ai Grok 4.6 · xAI Announcement (knowledge cutoff: "The knowledge cut-off date of Grok 4.6 is February 1, 2026." — docs.x.ai)

Performance — Matches GPT-5.6 Sol on Intelligence Index

From xAI's official evals table. Post-training improvements over Grok 4.5 across coding, agents, and knowledge work.

  • AA Intelligence Index: 61 — matches GPT-5.6 Sol Max (61), above Grok 4.5 High (56). Fable 5 Max is 62.
  • GDPVal-AA v2: 1753 — large knowledge-work Elo gain over Grok 4.5 (1526).
  • CursorBench v3.2: 69.9% — above GPT-5.6 Sol (67.2%).
  • DeepSWE v1.1: 65.9% — long-horizon software engineering, up from Grok 4.5 (54%). GPT-5.6 Sol is 73%.
  • FrontierCode v1.1 (Extended): 61.3% — production code quality, above Grok 4.5 (56.6%).
  • APEX-Agents: 57.5% — up over 10 pts from Grok 4.5 (47.1%).
  • AA-Briefcase: 1577 — large gain over Grok 4.5 (1313).

xAI positions Grok 4.6 as "our flagship model for code and everything else: agentic tool calling, minimal hallucinations, configurable reasoning." Third-party scores are best self-reported/public values; test conditions may differ.

Sources: xAI: Introducing Grok 4.6 (evals table)

Competitor Price Comparison

$2/$6 is among the cheapest in the flagship tier. All prices in USD per 1M tokens (standard, short context).

ModelProviderInputOutputvs Grok 4.6Notes
Grok 4.6xAI$2.00$6.00Baseline
GPT-5.6 SolOpenAI$5.00$30.00Grok cheaper60% cheaper input, 80% cheaper output
Claude Opus 5Anthropic$5.00$25.00Grok cheaper60% cheaper input, 76% cheaper output
GPT-5.6 TerraOpenAI$2.00$12.00Same inputGrok: half the output cost
Gemini 3.1 Pro PreviewGoogle$2.00$12.00Same inputGrok: half the output cost
Claude Sonnet 5Anthropic$2.00$10.00Same inputGrok: 40% cheaper output (Sonnet 5 intro price thru 8/31)
Gemini 3.7 FlashGoogle$0.75$3.75Gemini cheaperIntro price (thru 12/31). Flash tier
Grok 4.5xAI$2.00$6.00SamePrevious gen. Cached input $0.30

* Grok 4.6 is low-priced for the flagship tier, but lighter models (GPT-5.6 Luna $0.20/$1.20, Gemini 3.1 Flash-Lite $0.25/$1.50) serve a different, high-throughput niche. Grok 4.6 targets high-difficulty coding and agentic work.

Sources: official pricing pages from each provider (verified 2026-08-17). See full pricing comparison. Gemini 3.7 Flash details →

Ideal Use Cases

Long-running agents, high-difficulty coding, and knowledge work.

💻

Coding & Code Review

Strong CursorBench/DeepSWE scores. Production code generation, refactoring, review.

🤖

Long-Running Agents

Multi-step research, analysis, and cross-codebase work, with emergent self-verification.

📊

Knowledge Work & Research

Unfamiliar-domain research and synthesis. Strong GDPVal-AA / AA-Briefcase.

🎨

Interactive & Visual Builds

Turn a product idea into app structure + visual language in a first pass.

🛠️

Tool Use & Function Calling

Structured outputs, reasoning_effort control, MCP integration for agentic automation.

🔍

RAG & Document Search

500K context + collections search ($2.50/1k calls) for large-document Q&A.

When to Choose Grok 4.6

  • Flagship intelligence at $2/$6: Comparable-tier intelligence index at a fraction of GPT-5.6 Sol ($5/$30) or Claude Opus 5 ($5/$25). Strong cost-per-capability.
  • Watch the 200K wall: Long agent trajectories that exceed 200K tokens jump to $4/$12. Control cost with cached input ($0.50) and reasoning_effort.
  • Lightweight workloads elsewhere: For high-throughput, low-unit-cost work, consider GPT-5.6 Luna ($0.20/$1.20) or Gemini 3.1 Flash-Lite ($0.25/$1.50).

Estimate Your Actual Costs

Compare token costs for Grok 4.6 and all major models. Just enter your token counts.

🧮 Open Token Calculator →

📊 Full Pricing Comparison · ⚡ Compare with Gemini 3.7 Flash · 🗓️ Release Timeline · 🔗 xAI Official Pricing

⚠️ Disclaimer

  • Pricing verified against docs.x.ai on 2026-08-17. Subject to change without notice. Always check official pages before contracting.
  • This site is informational only. We are not affiliated with or endorsed by any provider.
  • Benchmark scores are xAI's published values. Your mileage may vary.