Claude Opus 4.6 (Thinking) completely unusable — persistent "high traffic" errors all afternoon, hundreds of retries with 0% success

[qin_yi]

I’ve been experiencing persistent capacity errors when using Claude Opus 4.6 (Thinking) in Antigravity IDE. The error message is:

“Our servers are experiencing high traffic right now, please try again in a minute.”

The attached screenshot shows just one example — the agent retried 5+ times before terminating with “Agent terminated due to error.” But in reality, I’ve been experiencing this issue the entire afternoon, retrying hundreds of times across multiple conversations with a 0% success rate. Claude Opus 4.6 (Thinking) has been completely unusable for the whole day, not just a single session.

I’m an Ultra subscriber paying for premium access, and getting zero successful completions over an entire afternoon is unacceptable. In the Asia-Pacific timezone (afternoon hours), the model is effectively non-existent.

Questions:

1. Is there a known capacity issue with Claude Opus 4.6 (Thinking) on Google Cloud?

2. Are there plans to improve availability for Ultra subscribers?

3. Is there any recommended workaround beyond switching to a less capable model?

Environment: Antigravity IDE, Ultra plan, Claude Opus 4.6 (Thinking), Asia-Pacific region

Thanks for any guidance.

[Pratik_Patil]

same error

Trajectory ID: d5438111-9e63-4602-
Error: agent executor error: could not convert a single message before hitting truncation
(1) attached stack trace
– stack trace:
| google3/third_party/gemini_coder/framework/executor/agentexecutor/agentexecutor.(*AgentExecutor).Run
| third_party/gemini_coder/framework/executor/agentexecutor/agentexecutor.go:312
| google3/third_party/jetski/cortex/cortex.(*CascadeManager).executeHelper.func1
| third_party/jetski/cortex/cascade_manager.go:2027
| google3/third_party/jetski/cortex/utils/background.(*Pool).Go.func2
| third_party/jetski/cortex/utils/background.go:121
| […repeated from below…]
Wraps: (2) agent executor error
Wraps: (3) attached stack trace
– stack trace:
| google3/third_party/jetski/cortex/chatconverters/chatconverters.init
| third_party/jetski/cortex/chatconverters/trajectory_chat_converter.go:47
| runtime.doInit1
| third_party/go/gc/src/runtime/proc.go:8138
| runtime.doInit
| third_party/go/gc/src/runtime/proc.go:8105
| runtime.main
| third_party/go/gc/src/runtime/proc.go:261
| runtime.goexit
| third_party/go/gc/src/runtime/asm_amd64.s:1774
Wraps: (4) could not convert a single message before hitting truncation
Error types: (1) *withstack.withStack (2) *errutil.withPrefix (3) *withstack.withStack (4) *errutil.leafError

[Saif_Ali3]

same here can’t use any model at all!

[Marwane_Bouziane]

same here, I also can’t use any model at all!

[ZDust]

All models (including Claude, Sonnet, Pro, and non-Pro) are currently not responding.