xAI cuts Grok 4.3 prices by 40-60%, adds creative agent mode
xAI released Grok 4.3 on May 2, a developer-focused model with steep price reductions and a new agent mode for creative projects. Input costs dropped 40 percent and output costs fell 60 percent compared to the previous version.
The model runs at 100 tokens per second with a one-million-token context window. Pricing sits at $1.25 per million input tokens and $2.50 per million output tokens, with a knowledge cutoff of December 2025.
What Grok 4.3 can do
Grok 4.3 handles web search, X search, Python code execution, and file search autonomously. It can generate Excel files, PDFs, and PowerPoint decks without human intervention.
Reasoning is now built in by default-the model "thinks" before answering every request. Reasoning tokens are billed at the same rate as regular output tokens.
The model is available through OpenRouter, the xAI API, and the Hermes agent.
Performance versus cost
According to Artificial Analysis, Grok 4.3 scored 53 on the Intelligence Index. That's four points above the previous Grok 4.20 but trails OpenAI's GPT-5.5 (60 points) and Anthropic's Claude Opus 4.7.
A full benchmark run costs $395 for Grok 4.3, compared to $3,959 for GPT-5.5 and $4,811 for Claude Opus 4.7. The model lands on what Artificial Analysis calls the Pareto frontier-where performance and cost align best.
On GDPval-AA, a benchmark measuring real-world knowledge work tasks, Grok 4.3's Elo score jumped 321 points to 1,500. It now ranks ahead of Google's Gemini 3.1, though GPT-5.5 still leads by 276 Elo points.
Results vary on other benchmarks. Val's AI ranks Grok 4.3 first on contract law and corporate finance tasks but found it struggles with general coding and harder math problems, placing 13th. Andon Labs reported the model sometimes sits idle on autonomous agent tasks instead of taking action.
New Imagine Agent Mode for creatives
xAI launched Agent Mode for Grok Imagine in beta on the web interface. Rather than responding to single prompts, the mode handles longer creative projects by planning, generating, editing, and revising content in an open workspace.
The tool works for one-minute movies, manga sets, and product stories. Users can turn it on in the input field at the bottom left of the Grok web interface. A paid account is required.
This feature targets creatives who need to manage multi-step production workflows. For teams working on visual or narrative projects, the agent mode offers a way to maintain consistency across revisions without resetting context between steps.
Learn more about Grok (from xAI) and AI for Creatives.
Your membership also unlocks:
