Less than three months after releasing its first voice model, xAI just shipped a substantially better one. Grok Voice Think Fast 2.0, announced July 29, 2026, improves on every dimension that matters for enterprise voice AI: speed, accuracy, language coverage, and intelligence.

This is not a minor update. The numbers suggest a model that closes the gap with purpose-built voice AI leaders while running at a price that keeps the economics competitive.

What Changed in Think Fast 2.0

Speed: Time to first audio dropped from 1.25 seconds to 0.70 seconds. That half-second difference matters in voice conversations more than it does in text-based interfaces. A caller hears the pause. A customer on a support call that feels sluggish hangs up or gets frustrated. Sub-second response is the threshold where voice AI starts to feel natural rather than robotic.

Accuracy: Think Fast 2.0 delivers a 1.4x improvement in transcription accuracy over Think Fast 1.0. Against widely used alternatives like Deepgram Nova 3 and ElevenLabs Scribe v2, xAI claims 1.5x to 2.0x better accuracy. In noisy environments, the gap reportedly widens to roughly 10x. That matters in practice: call centers, field service, retail, and any scenario where callers are not sitting in a quiet room with a headset.

Intelligence: The model handles more complex reasoning while cutting the tokens used for reasoning by approximately 60 percent. The practical effect is better handling of nuanced requests, more reliable tool use, and improved conversational ability, without proportional cost increases.

Languages: Think Fast 2.0 supports more than 25 languages and is built to handle varied accents across all of them. For businesses operating across multiple regions or serving multicultural customer bases, this removes one of the more common reasons to rule out a single voice AI vendor.

Real Enterprise Results

xAI did not just release benchmark numbers. The company ran an A/B test on Starlink’s phone service using Think Fast 2.0 versus the previous model. The results showed a meaningful increase in both sales conversion rate and support containment rate.

That distinction is worth noting. Most voice AI vendors publish latency numbers and accuracy benchmarks. Fewer release the downstream business outcomes. The fact that xAI is sharing conversion and containment data suggests they are positioning this model for enterprise buyers who care about ROI rather than just model specs.

Pricing and Availability

Think Fast 2.0 is available now in the xAI API and Voice Agent Builder at $0.08 per minute. Existing users who have specified “grok-voice-latest” in their integrations will be automatically upgraded to Think Fast 2.0 starting August 5, 2026, with no action required.

What This Means for Business

A few months ago, the voice AI market had a handful of serious vendors and a clear tier structure between the leaders and the challengers. xAI’s rapid iteration is compressing that structure.

Think Fast 2.0 is not the fastest, most accurate, or cheapest voice AI on the market in every category. But it has closed the gap significantly in two model generations, and xAI has the compute infrastructure and funding to keep iterating quickly. The pace of improvement is itself a signal worth tracking.

For businesses evaluating voice AI right now, this creates a specific dynamic: the capabilities you can buy today are meaningfully better than what was available six months ago, and six months from now they will likely be better still. Waiting for “good enough” keeps getting more expensive in opportunity cost.

For existing voice AI deployments: If you are running a phone agent or voice-enabled service, Think Fast 2.0’s accuracy improvements in noisy environments are worth testing against your actual call data. A 1.5x accuracy gain on real-world calls, if it holds, translates directly into fewer misunderstood requests and fewer escalations to human agents.

For businesses still evaluating: The fact that a major tech player with infrastructure at scale is now competing aggressively in enterprise voice AI is good news for buyers. You have more leverage in vendor conversations than you did a year ago, and the economics of deploying voice AI continue to improve.

The voice AI market is moving fast. The businesses that get into deployment now, even imperfectly, are accumulating the real-world experience that turns a technology pilot into a genuine operational advantage.

---

Enterprise DNA’s Omni Voice service designs, deploys, and refines enterprise voice AI employees for businesses ready to move from evaluation to production. Book a discovery call to explore what voice AI can actually do for your operations.