Grok 4.3 – Speed and Intelligence at Fraction of the Cost

Trusted by users from 10,000+ companies

Additional Capabilities in Grok 4.3

Additional Capabilities in Grok 4.3 A broader set of improvements that extend Grok 4.3's usefulness across development, research, and enterprise workflows.

Always-on Reasoning

Reasoning is active by default on every request and cannot be disabled, ensuring consistent depth on complex and ambiguous tasks without configuration.

Code Execution

Ships with a built-in environment that can write, run, and produce files, enabling end-to-end code workflows without external tooling.

Agent Library

A new client-side agent library gives developers structured access to Grok 4.3's agentic capabilities for building and deploying custom workflows.

Frequently Asked Questions

Discover more about Grok 4.3 through common user queries online.

Voice Cloning

Custom Voices, xAI's new voice cloning suite, is available alongside Grok 4.3, priced at $4.20 per million characters with natural prosody controls.

Speech-to-Text API

Now generally available with transcription across 25 languages, batch and streaming modes, and multi-speaker diarisation with word-level speaker IDs.

Text-to-Speech API

The same voice stack powering Grok Voice and Tesla vehicles is now accessible via API, with inline speech tags for prosody control at scale.

Tiered Context Pricing

Requests up to 200K tokens are billed at standard rates. Beyond that, a higher tier applies, giving teams control over cost based on actual context usage.

OpenAI-compatible API

Grok 4.3 is fully accessible via OpenRouter and compatible with the OpenAI SDK, making it a low-friction drop-in for existing API integrations.

Grok Computer Integration

Tighter integration with Grok Computer which enables Grok 4.3 to take direct action on software tasks beyond the chat interface.

Strong Benchmark Results

Grok 4.3 sits on the Pareto frontier for intelligence versus cost, scoring higher while costing less to run than its predecessor.

Real-world Knowledge Work

Grok 4.3 scores an ELO of 1500 on GDPval-AA, a 321-point jump over its predecessor. It’s ahead of multiple frontier models on the benchmark designed to measure performance on economically valuable knowledge tasks.

Unreal Performance vs Cost

Despite scoring higher on the Artificial Analysis Intelligence Index, Grok 4.3 costs around 20% less to run the full benchmark suite than Grok 4.20. More capable and cheaper to operate among top-performing competitor models.

Sharper Factual Recall

Grok 4.3 gains 8 points on AA-Omniscience Accuracy over Grok 4.20, reflecting a measurable improvement in factual retrieval across knowledge-intensive tasks. For legal research, technical analysis, scientific reasoning.

Long-document Analysis at Scale

With a 1 million token context window and no output token limit, Grok 4.3 can take in entire legal case histories, financial filings, research libraries, or large codebases in a single session.

Video and Documents from a Single Workflow

Grok 4.3 accepts text, images, and video natively, and produces downloadable PDFs, fully formatted spreadsheets, and presentation decks directly in conversation. Early testers report outputs not needing additional editing.

Parallel Agents on Complex Tasks

For complex planning tasks, long-horizon research, or multi-part simulations, this architecture compresses timelines that would otherwise require sequential execution across hours.

Grok 4.3: Built for Long Documents and Agentic Work

Additional Capabilities in Grok 4.3

Always-on Reasoning

Code Execution

Agent Library

Frequently Asked Questions

What is Grok 4.3?

How much does Grok 4.3 cost via API?

Who can access Grok 4.3?

Can Grok 4.3 process video?

What is the 16-Agent Heavy system?

What is Grok 4.3's context window?

Can reasoning be turned off in Grok 4.3?

How does Grok 4.3 compare to Grok 4.20?

Does Grok 4.3 support persistent memory?