← Back to API models

Grok 4.3

by xAI · reasoning workhorse · replaced Grok 4

xAI's mid-tier reasoning model and the direct successor to the retired Grok 4 — a bigger context window at a lower price. Tiered pricing: $1.25/$2.50 per 1M under 200K tokens, $2.50/$5 per 1M at 200K tokens and above.

Input <200K $1.25 / 1M
Output <200K $2.50 / 1M
Context 1M tokens
≥200K tokens $2.50 / $5
xAI Console ↗ Updated July 27, 2026
§ API pricing

Per-token rates, with a tier break at 200K.

Input <200K
$1.25/1M tokens
Standard tier
  • Applies while the prompt stays under 200K tokens
  • Well under Grok 4.5's $2 flagship rate
  • Cheaper than the retired Grok 4's $3
Output <200K
$2.50/1M tokens
Standard tier
  • 2× the standard-tier input rate
  • A sixth of the retired Grok 4's $15
  • Includes reasoning tokens
≥200K tokens
$2.50/$5 per 1M
Long-context tier
  • Kicks in once the prompt reaches 200K tokens
  • Applies to the full request, not just the overage
  • Double the standard input and output rates
Context
1Mtokens
Window
  • Nearly 4× the retired Grok 4's 256K
  • Function calling supported
  • Web search, X search, code exec

Grok 4.3 bills $1.25 / 1M input and $2.50 / 1M output for prompts under 200K tokens. Once a request reaches 200K tokens, the entire request is billed at $2.50 / 1M input and $5 / 1M output.

Why Grok 4.3 exists

Grok 4 no longer appears in xAI's official API pricing table — it has been fully retired and superseded by Grok 4.3, which now fills the reasoning-workhorse slot underneath the Grok 4.5 flagship. Grok 4.3 undercuts the old Grok 4 rate on both input and output while shipping a much larger context window: 1M tokens versus Grok 4's 256K. Anyone routing traffic to the old model ID should move to Grok 4.3, which is now the cheaper, more capable option in that tier.

Pricing is tiered, matching the pattern set by Grok 4.5: $1.25 input / $2.50 output per 1M tokens while a request stays under 200K tokens, rising to $2.50 / $5 per 1M — applied to the whole request — once it crosses that threshold. That gives budget-conscious teams a predictable way to control cost by keeping prompts under the 200K mark.

Capabilities

Grok 4.3 is positioned as a solid, general-purpose reasoning model: capable tool use, function calling, and code assistance without the premium price of the flagship tier. It's the model to reach for when a task clearly doesn't need Grok 4.5's top-end reasoning but is still too demanding for a budget or code-specialist model. Real-time web and X search carry over from the rest of the Grok lineup, and the 1M-token window comfortably handles long documents and multi-file codebases.

The honest trade-off: it won't match Grok 4.5 on the hardest reasoning and agentic benchmarks. For frontier-grade tasks, step up to the flagship; for routine reasoning, drafting, and moderate-complexity coding, Grok 4.3 is the more economical fit.

Typical use cases

  • General-purpose reasoning and drafting at a mid-tier price
  • Moderate-complexity coding and debugging assistance
  • Long-document analysis up to 1M tokens
  • Tool-use and function-calling pipelines that don't need the flagship
  • Migrating workloads off the retired Grok 4 at a lower cost

Sibling and rival comparison

ModelInput / 1MOutput / 1MContext
Grok 4.3 (<200K)$1.25$2.501M
Grok 4.3 (≥200K)$2.50$51M
Grok 4.5$2$6500K
Grok Build 0.1$1$2256K
Claude Sonnet 5$3$151M
Gemini 3.5 Flash$1.50$91M

Grok 4.3 sits neatly between xAI's budget code model, Grok Build 0.1, and its Grok 4.5 flagship — cheaper than Sonnet 5 on output by a wide margin, while carrying a full 1M-token window. It's the model to default to when a task doesn't clearly need flagship-tier reasoning.

← See all Grok / xAI plans