ACTION REQUIRED — AUGUST 5 DEADLINE (TOMORROW)
● What happens August 5: grok-voice-latest alias moves from Think Fast 1.0 to Think Fast 2.0
● Price change: $0.05/min → $0.08/min (+60%) — happens automatically for grok-voice-latest users
● To upgrade: Do nothing — grok-voice-latest auto-upgrades
● To stay on v1.0: Pin grok-voice-think-fast-1.0 before August 5
● AA Speech-to-Speech Quality Index: 2.0 scores 82.9% vs 1.0 at 75.7%
● vs GPT-Realtime-2.1: 82.9% vs 79.1% — Think Fast 2.0 leads
● vs Gemini 3.1 Flash: 82.9% vs 69.5%
● Prompt compatibility: xAI says existing prompts work without changes
● Voice Agent Builder: Confirm which model ID it pins after August 5 before assuming price
The Decision: Pin v1.0 or Accept 2.0?
According to SpaceXAI's official announcement, grok-voice-latest will automatically move to Think Fast 2.0 on August 5 with no action required to upgrade. The 60% cost increase — from $0.05 to $0.08 per minute — is the variable that makes this decision worth reviewing before the automatic switch happens. As Digital Applied's analysis notes, "anything pointing at the alias moves from a $0.05 to a $0.08 per-minute audio rate on August 5 without a deployment, a changelog entry, or a line in anyone's sprint." A support line running a few thousand minutes per month absorbs a 60% increase in its audio line item silently if no action is taken.
The independent benchmark evidence favours upgrading. Per Testing Catalog's coverage, Artificial Analysis scored Think Fast 2.0 at 82.9% on its Speech-to-Speech Quality Index, up from 75.7% for version 1.0, ahead of GPT-Realtime-2.1 at 79.1% and Gemini 3.1 Flash at 69.5%. A 7.2-point improvement on a quality index, combined with 0.70-second first audio response and 10× better transcription in noisy environments, is a material capability upgrade — not a cosmetic one. For most voice agent deployments, the accuracy and speed improvements likely justify the 60% cost increase. But that judgment depends on your specific call types, volumes, and the revenue or containment value each call represents.
What Changed in Think Fast 2.0 vs 1.0
| Metric | Think Fast 1.0 | Think Fast 2.0 |
|---|
| AA STS Quality Index | 75.7% | 82.9% |
| First audio response | 1.25 seconds | 0.70 seconds |
| Transcription vs competitors | Baseline | 1.5-2× better (10× in noise) |
| Reasoning tokens | Baseline | ~60% fewer |
| Price per minute | $0.05 | $0.08 (+60%) |
Three Operational Notes Before August 5
1. Pin explicitly in production, float in staging. As ExplainX.ai's developer guide recommends, the production rule is: pin a versioned ID (grok-voice-think-fast-2.0 or grok-voice-think-fast-1.0). Let grok-voice-latest float only in staging environments. Never use an unversioned alias in production — the August 5 event is exactly why.
2. Run regression tests before August 5. Per We0.ai's assessment, xAI says most applications should see improved performance without prompt changes — but "timing, tool use, turn-taking, and response style can change between model versions." Run your call scripts and tool schemas on 2.0 before August 5.
3. Check Voice Agent Builder pricing separately. Digital Applied flags that Voice Agent Builder advertised $0.05/min at its July 1 launch. Whether the platform price moves when the default model flips on August 5 is unconfirmed. Pull a fresh usage export after August 5 rather than assuming Builder and raw API share one price.
Sources: SpaceXAI official announcement · Testing Catalog / Artificial Analysis benchmark · Digital Applied · ExplainX.ai developer guide · We0.ai · Related: Full Think Fast 2.0 launch review →