Grok 4.1 Is Being Removed from Google Cloud: The Routing Migration Every GEAP Operator Must Finish Before August 20
Google Cloud's Gemini Enterprise Agent Platform shuts down Grok 4.1 Fast on August 20, 2026. API requests then return 400 — no fallback, no grace period. If your routing layer points to xai/grok-4.1-fast on GEAP, here is the migration decision matrix.
Archive item produced with AI assistance from the cited source and published without individual review. Editor of record: Joe Werner.

Google Cloud moved quietly on July 8, 2026: the Gemini Enterprise Agent Platform (GEAP) announced that the entire Grok 4.1 model family is deprecated and will be permanently shut down on August 20, 2026. After that date, API requests to xai/grok-4.1-fast-reasoning or xai/grok-4.1-fast-non-reasoning on GEAP will return a 400 error with no fallback.
If you route any Grok traffic through Google Cloud rather than directly through the xAI API, the migration timeline is different — and the cost math is different too.
What happened
On July 8, GEAP added a deprecation notice for the Grok 4.1 model family across its Model as a Service (MaaS) endpoints:
- Models affected:
xai/grok-4.1-fast-reasoningandxai/grok-4.1-fast-non-reasoning - Deprecation date: July 8, 2026
- Hard shutdown: August 20, 2026 at 00:00 UTC
- What happens after shutdown: Requests to the
https://aiplatform.googleapis.com/v1/projects/<your-project>/locations/global/endpoints/openapi/chat/completionsendpoint using Grok 4.1 model IDs will fail with a400error
Google's guidance is to migrate to newer xAI models available in GEAP — specifically Grok 4.2 or Grok 4.3 — or to choose an alternative from the Model Garden.
This is separate from the xAI direct API retirement (which was May 15, 2026) and from the Oracle Cloud Infrastructure retirement (August 15, 2026, with deprecation effective May 15). Cloud provider timelines for the same model family can diverge by months.
Why it matters for AI engineering teams
Grok 4.1 Fast was positioned as xAI's best agentic tool-calling model for real-world workloads — customer support pipelines, deep research agents, and high-frequency tool-invocation tasks. Teams that adopted it through GEAP specifically for Google Cloud infrastructure integration (IAM, billing, VPC-SC, Workspace) now face a forced migration with 43 days of notice.
The operator-facing pain points are real:
Cost step-up. Community reports indicate the nearest replacement — Grok 4.3 on GEAP — carries pricing approximately 5× higher than Grok 4.1 Fast was at deprecation. For high-volume tool-calling workloads, that is a significant unit economics shift that may require re-evaluating which model tier is appropriate versus routing to a cheaper alternative entirely.
Routing configuration changes. GEAP uses an OpenAI-compatible endpoint with xai/ prefixed model names. Teams using Cursor, Claude Code, or other OpenAI-compatible clients pointed at GEAP need to update the model parameter in their config — not just the base URL. Config-driven routing frameworks need a model-id remap.
No auto-forwarding. Unlike some deprecations where the old model ID silently aliases to a newer version, GEAP will return a 400 error after August 20. There is no soft transition.
Multi-cloud teams have fragmented timelines. If your organization uses both GEAP and Oracle Cloud Infrastructure for Grok access, the retirement dates differ (OCI: August 15; GEAP: August 20). A single "Grok 4.1 is retired" deployment change may not be sufficient — each cloud endpoint needs an independent audit.
The router/operator angle
This deprecation exposes a structural issue with cloud-provider-hosted third-party models: deprecation SLAs are set by the cloud provider, not the original model vendor.
When xAI retired Grok 4.1 Fast from its direct API on May 15, operators who accessed Grok through Google Cloud or Oracle had additional runway — but only if they were tracking GEAP and OCI release notes independently. Operators who assumed "xAI API retirement = everywhere retirement" had a false sense of safety; operators who assumed "GEAP follows xAI's timeline" had no visibility into the August 20 deadline until they happened to check GEAP release notes.
The migration decision matrix for affected teams:
| Current model (GEAP) | Replacement option | Cost impact | Routing change needed |
|---|---|---|---|
xai/grok-4.1-fast-reasoning | xai/grok-4.3 | ~5× higher per token | Update model param |
xai/grok-4.1-fast-non-reasoning | xai/grok-4.2 | Moderate increase | Update model param |
| Either | Switch to Gemini 3.5 Flash on GEAP | Lower for most workloads | Update model + re-eval tool schemas |
| Either | Route Grok directly via xAI API | Depends on agreement | New base URL + auth config |
What to audit now:
- Grep your config files. Search for
grok-4.1,grok-4.1-fast,grok-4.1-fast-reasoning,grok-4.1-fast-non-reasoningacross all routing config, environment variables, and SDK initialization code. - Check which endpoint base URL you use. If
base_urlpoints toaiplatform.googleapis.com, you are on GEAP and affected. If it points toapi.x.ai, you are on the direct API (already past deadline). - Cost-model the replacement. For high-volume tool-calling workloads, the 5× cost jump to Grok 4.3 may make a lateral move to Gemini 3.5 Flash or DeepSeek V4 Pro more economical than a like-for-like model swap.
- Update monitoring. Add a
400error rate alert on the affected model IDs with a hard trigger threshold before August 20.
What TheRouter users should watch or try
If you route Grok requests through a gateway that terminates at GEAP, the model ID update must happen before August 20. The same OpenAI-compatible model field that specifies the Grok variant needs to change to xai/grok-4.2 or xai/grok-4.3 — or the route needs to be redirected to a different provider endpoint entirely.
For teams managing multiple cloud provider routes, this is a good forcing function to audit your model ID → provider endpoint mapping table. A gateway that abstracts provider selection (routing grok-4.1-fast-reasoning semantically to whichever active endpoint is available) is better positioned to survive this kind of cloud-provider-specific retirement than one that hardcodes the full GEAP model identifier in application code.
Check the GEAP migration guide for the official model substitution table. If you run Grok on Oracle Cloud Infrastructure, note your deadline is August 15 — five days earlier.

DashScope Qwen Native Model Retirement October 2026: The Routing Audit Your Team Cannot Skip
Alibaba Cloud Model Studio retires qwen-turbo, qwq-plus, qvq-max, qwen-vl-max, qwen-coder-turbo, qwen-tts, and the full paraformer ASR stack on October 10. QPM throttling has already started — here is the routing audit checklist.

Alibaba Bailian Retires DeepSeek, Kimi, MiniMax, and GLM Models on October 10: The Multi-Provider DashScope Routing Audit
Alibaba Bailian retires DeepSeek V3/R1, Kimi K2, MiniMax M2.1, and GLM 4.6/4.7 proxies on October 10, 2026 — plus qwen-coder-plus and dozens of Qwen3 variants. QPM throttling has already begun; audit your DashScope model IDs now.

Gemini 3.5 Flash Computer Use API Is Now a Built-in Tool: The Routing Decision That Changes Your Agentic Stack
Gemini 3.5 Flash now ships computer use as a built-in tool alongside Search and Maps grounding — no separate model required. That consolidation changes how operators route browser and desktop automation workloads.