Grok Retired Models List: What Stopped Working on May 15, 2026

May 15, 2026 marked a significant breakpoint in the world of AI models, particularly impacting users of Grok and its affiliated ecosystem including notable players like xAI and SpaceXAI. This post dives into the https://technivorz.com/is-grok-business-billed-monthly-and-can-i-cancel-seats/ detailed list of Grok model retirements—including Grok 4.1 Fast Retired, Grok 4 Fast Retired, and Grok 3 Retired—and breaks down the 2026 pricing changes and their implications. We'll also dissect consumer tiers versus X-bundled tiers, explain the math behind annual billing discounts, and highlight what remains uncertain about free tier limitations.

Grok Model Retirements as of May 15, 2026

The first critical event in 2026 was the official retirement of several key Grok models. Users dependent on these models found them inaccessible starting on May 15, 2026—a hard cutoff point. Here is a quick summary:

Grok 4.1 Fast Retired: Once popular for its speed and nuanced responses, Grok 4.1 Fast was fully deprecated and ceased operation on the breakpoint date.
Grok 4 Fast Retired: The base Grok 4 Fast, a cornerstone in many AI-powered services, was discontinued simultaneously, forcing users to upgrade or migrate.
Grok 3 Retired: The previous generation Grok 3 model, already phased out in many tiers, was fully retired on this date.

These retirements affect developers integrating APIs for per-token billing and tools using cached input tokens pricing, which are common in many Grok-powered SaaS applications.

Timeline and Context for 2026 Pricing Changes

The May 15, 2026 breakpoint came alongside a broader restructuring of pricing models from Grok and its extended network, including xAI’s AI solutions and SpaceXAI’s specialized offerings. Here's an overview timeline:

Early 2026: Grok announces retirement timelines for legacy models with a warning about affected API endpoints and subscription tiers.
Mid-May 2026 (May 15): Official deactivation of Grok 3, Grok 4 Fast, and Grok 4.1 Fast models.
Post-May 15, 2026: New pricing and tiers roll out, emphasizing bundles with X-services and enhanced consumer vs. X-bundled segmentation.

This was not an isolated tech shift—it reflects a broader consolidation of AI services that integrate Grok’s capabilities with next-gen offerings from the xAI and SpaceXAI brands. It also forces users to reevaluate subscriptions to keep pace with discontinued support.

Consumer Tiers vs. X-Bundled Tiers: What's Changed?

One of the thornier issues accompanying the model retirements is the segmentation of pricing and feature sets into consumer tiers and X-bundled tiers. Here’s what you need to know:

Consumer Tiers: These tiers mainly target individual developers, small teams, and startups. They provide straightforward API token pricing and access to core Grok-powered models, but with strict usage limits and fewer bundled benefits.
X-Bundled Tiers: Primarily aimed at enterprise customers and power users leveraging multiple AI tools—often branded under or co-developed with xAI and SpaceXAI. These include bundled access, priority support, and optimized token caching for priced efficiency.

Consumers on these different tiers face distinct challenges. For example, the free tiers under consumer plans have grown more restrictive, with unknown limits on real-time token usage, while X-bundled tiers offer improved caching strategies that reduce billed tokens and improve cost-efficiency.

Annual Billing Discount Math: The Real Cost Behind SuperGrok Lite

Pricing transparency is often murky following model retirements and re-tierings. Grok's new plans use many annual+monthly billing combinations, leading to confusion. Let’s clarify with the example of SuperGrok Lite:

Plan Cost per Month Annual Cost Effective Monthly Cost (Annual) SuperGrok Lite $10/month $100/year $8.33/month

The $100/year charge equates to $8.33/month effective cost, offering a 16.7% discount compared to a $10/month month-to-month subscription. This math is crucial since many users blindly trust rounded pricing or ignore these savings when choosing their billing cadence.

However, the discount only applies if you commit upfront. Canceling early or switching plans mid-year might negate the savings. Also, this example is for a simple consumer tier—X-bundled tiers can have more complicated math factoring in token caching and API call volume.

API Per-Token Billing and Cached Input Tokens Pricing

API pricing models are shifting from flat-rate to highly elastic per-token billing schemes, and Grok’s ecosystem exemplifies this trend. Two pricing dimensions deserve special attention:

Per-Token Billing: Charges are calculated on the actual tokens processed by API calls. Each token generated or consumed incurs a fee. This approach aligns costs closely with usage but penalizes inefficient or verbose API queries.
Cached Input Tokens Pricing: Grok and partners increasingly apply caching strategies that avoid re-billing tokens for repeated inputs within certain timeframes. This innovation benefits users with repeated prompt sequences or dataset queries, reducing unnecessary charges.

Users should note that cached tokens pricing often resides in X-bundled tiers or higher-end consumer tiers. Those on free or bare-bones plans might not see these cost savings, compounding the impact of model retirements forcing more frequent API calls.

Free Tier Limitations and Uncertainties

Free tiers have historically been vital for experimentation and small-scale development with Grok models. However, after May 15, 2026, several limitations surfaced:

Many retired models exit free tier availability entirely.
New limitations cap token counts per day or per session, often without transparent documentation.
Uncertainty remains around whether cached token pricing benefits propagate to free users or remain privileged in paid tiers.

Unfortunately, Grok's official communications have been vague, leaving some users unsure about their actual free-tier quotas or throttling thresholds. This opacity complicates evaluation of services for low-volume users and developers testing new features.

How xAI and SpaceXAI Factor Into the Grok Ecosystem

xAI and SpaceXAI, two rising AI companies linked to Grok’s parent ecosystem, have adopted Grok infrastructure in their offerings. Post-May 15, 2026, these companies have:

Embraced the new model tiers, especially leveraging X-bundled service options to optimize AI workloads.
Negotiated pricing that factors in cached input token benefits to enhance cost efficiency for demanding customers.
Implemented custom tools aimed at offsetting model retirements by introducing next-gen AI models bundled with Grok services.

This trend highlights how AI providers aim to future-proof their services through consolidation and layered service models, though it creates complexity for users juggling multiple subscriptions.

Summary and Recommendations

To recap:

Grok 4.1 Fast Retired, Grok 4 Fast Retired, and Grok 3 Retired models ceased functionality on May 15, 2026—a hard breakpoint forcing upgrades or migrations.
Pricing shifts introduced clearer but segmented consumer and X-bundled tiers, with significant differences in per-token billing and token caching strategies.
Annual billing discounts (e.g., SuperGrok Lite's effective $8.33/month vs $10 month-to-month) offer cost savings but require commitment.
Free tier limitations remain ambiguous, with increasing restrictions post-retirement.
xAI and SpaceXAI are leveraging Grok’s infrastructure in bundled offers aimed at enterprise and power users.

Final Caveats: Always sanity-check pricing details against official changelogs and monitor your API token consumption closely. Watch for model deprecation notices and avoid relying on retired APIs, especially around breakpoints like May 15, 2026.

For developers and small teams, the shift means weighing whether to pay for X-bundled plans—often more cost-effective for heavy usage—or stay lean on consumer tiers https://stateofseo.com/grok-business-pricing-at-30-per-seat-what-do-teams-actually-get/ with tight token economies. The math behind annual pricing, token caching, and API consumption is complex but crucial to optimize your AI costs moving forward.

Stay vigilant and maintain alternative workflows if your core models retire, since AI pricing and support will only grow more dynamic as we approach future breakpoints like August 12, 2026.