Grok 4.3 (xAI)

Grok 4.3 is positioned by xAI as a flagship release aimed squarely at agent-based and productivity-focused workloads, with particular emphasis on deep reasoning and autonomous tool use. Cloudflare's third-party catalog entry describes the model as having strong agentic tool calling with minimal hallucinations, and it documents the cataloged API limit that quick-info value window that lets the model ingest very large documents, codebases, or multi-turn agent traces in a single pass. The same catalog note confirms that Grok 4.3 accepts both text and image inputs, so it can handle mixed visual-and-textual prompts typical of retrieval-augmented agents, document understanding, and vision-grounded automation pipelines.

Microsoft's Foundry launch announcement reinforces the model's flagship positioning, describing Grok 4.3 as xAI's latest option for agent-based and domain-specific applications within a production-ready environment that emphasizes governance and cost efficiency. Cloudflare's listing shows practical developer surface area that supports this agentic framing: function calling for tool integration, structured outputs for reliable downstream parsing, and configurable reasoning effort with explicit none, low, medium, and high levels so teams can trade latency against answer quality per task. The Cloudflare entry also flags zero data retention, which matters for enterprise workflows handling sensitive inputs, and links directly to xAI's own documentation and terms for teams that need authoritative pricing and license details before deployment.

LLM Gatewayxai/grok-4-3grok

Open provider page Provider docs

Quick Info

Provider: LLM Gateway
Model key: xai/grok-4-3
Release date: Apr 30, 2026
Last updated: Apr 30, 2026
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $1.25
Output token cost: $2.50

Limits

Output tokens: 1,000,000 tokens
Context window: 1,000,000 tokens

Transparent token rates

Compare Grok 4.3 (xAI) pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about Grok 4.3 (xAI)

LLM Gateway

CoverageAnalysis

The kie.ai deep dive reports that Grok 4.3 surfaced publicly through an OpenRouter listing dated April 30, 2026, with a 1M-token context, configurable effort levels (none/low/medium/high, default low), and tiered pricing where requests above 200K total tokens get billed at a higher rate. Public pricing is listed at $1. Per Artificial Analysis, the article states the high-reasoning variant of Grok 4.3 measured 38 on the AA Intelligence Index v4.1 (rank 37/154) and roughly 152 output tokens per second, which conflicts with other candidates that cite 53, likely reflecting a different reasoning-effort variant rather than label clarity. A

LLM Gateway

CoverageAnalysis

Grok 4.3 shipped to xAI's public API on April 30, 2026 under the model id "grok-4.3," after entering beta on grok.com and the SuperGrok apps on April 17 and reaching general availability the week of May 4. The release is defined by a price drop rather than a spec upgrade: input falls roughly 40% and output roughly 60% Three notable capability changes ship with Grok 4.3: reasoning is now always-on by default with no separate toggle (billed at standard output rates), Grok 4.3 is the first xAI API model to natively accept video input processed through a vision encoder, and xAI simultaneously launched Custom Voices — a voice-cloning API

LLM Gateway

Coverage

Microsoft announced on May 13, 2026 that Grok 4.3, xAI's latest flagship model, is generally available on Microsoft Foundry, giving developers an enterprise-grade channel for deploying the model alongside safety controls, monitoring, and operational tooling. The post frames Grok 4.3 as purpose-built for agentic systems On Microsoft Foundry specifically, Grok 4.3 supports up to a 200,000-token context window for multi-turn reasoning and agent workflows — notably smaller than the 1M tokens claimed elsewhere — and delivers multimodal analysis across text, images, diagrams, and mixed data sources. The model also demonstrates strong web d

LLM Gateway

Coverage

xAI launched Grok 4.3 on the xAI API on May 5, 2026, as its fastest, most intelligent, and most cost-effective model to date. Key specifications: 1M-token context window, input pricing of $1.25 per million tokens (37% cheaper than Grok 4.20), output pricing of $2.50 per million tokens (58% cheaper than Grok 4.20), a sc The most significant benchmark improvement is on GDPval-AA, where Grok 4.3 scores an ELO of 1500 — up 321 points from Grok 4.20 0309 v2's 1179 — surpassing Gemini 3.1 Pro Preview, Muse Spark, GPT-5.4 mini, and Kimi K2.5 on real-world agentic task execution. xAI researcher Eric Jiang confirmed on X that a larger model t