URGENT: Claude Opus 4.6 Thinking - Repeated 503 "No capacity" errors on ULTRA tier

Hello,

I am an ULTRA subscriber paying premium pricing, and I cannot work.

For the past 20+ minutes, I have been receiving constant 503 UNAVAILABLE errors when trying to use Claude Opus 4.6 Thinking in Antigravity. This is unacceptable for a paid tier.

ERROR:
"No capacity available for model claude-opus-4-6-thinking on the server"

TRAJECTORY ID: 92dc8413-6c88-4df9-a845-a0c5239effa1
DATE/TIME: 2026-02-09 08:50 - 09:08+ EST (ongoing)
FREQUENCY: Every single request fails

I understand capacity constraints exist, but ULTRA subscribers should have reserved capacity or priority queue access over free/lower tiers. That's the entire point of paying premium.

If paying customers cannot reliably access premium models, what exactly are we paying for?

REQUESTS:
1. Immediate explanation of what's causing this outage
2. Reserved capacity or priority access for Ultra subscribers
3. SLA guarantees for paid tiers - if the model is unavailable, we need fallback options that work
4. Compensation/credit for lost productivity during outages

I chose Ultra specifically for reliability on demanding models. Please reserve capacity for those of us paying the premium price.

Awaiting urgent response.

John

Trajectory ID: 92dc8413-6c88-4df9-a845-a0c5239effa1
Error: model unreachable: UNAVAILABLE (code 503): No capacity available for model claude-opus-4-6-thinking on the server: UNAVAILABLE (code 503): No capacity available for model claude-opus-4-6-thinking on the server: model api cannot be reached
(1) attached stack trace
– stack trace:
| google3/third_party/gemini_coder/framework/generator/generator.(*PlannerGenerator).generateWithModelOutputRetry
| third_party/gemini_coder/framework/generator/planner_generator.go:194
| google3/third_party/gemini_coder/framework/generator/generator.(*PlannerGenerator).Generate
| third_party/gemini_coder/framework/generator/planner_generator.go:95
| google3/third_party/gemini_coder/framework/executor/executor.(*Executor).Execute
| third_party/gemini_coder/framework/executor/executor.go:303
| google3/third_party/jetski/cortex/cortex.(*CascadeManager).executeHelper.func1
| third_party/jetski/cortex/cascade_manager.go:1558
| […repeated from below…]
Wraps: (2) secondary error attachment
| UNAVAILABLE (code 503): No capacity available for model claude-opus-4-6-thinking on the server: UNAVAILABLE (code 503): No capacity available for model claude-opus-4-6-thinking on the server
| (1) tags: map[stream_receive_count:0 streaming_duration:0s]
| Wraps: (2) attached stack trace
| – stack trace:
| | google3/third_party/gemini_coder/framework/generator/generator.(*streamResponseHandler).processStream
| | third_party/gemini_coder/framework/generator/stream_handler.go:338
| | google3/third_party/gemini_coder/framework/generator/generator.(*PlannerGenerator).attemptGenerate
| | third_party/gemini_coder/framework/generator/planner_generator.go:437
| | google3/third_party/gemini_coder/framework/generator/generator.(*PlannerGenerator).generateWithAPIRetry
| | third_party/gemini_coder/framework/generator/planner_generator.go:278
| | google3/third_party/gemini_coder/framework/generator/generator.(*PlannerGenerator).generateWithModelOutputRetry
| | third_party/gemini_coder/framework/generator/planner_generator.go:154
| | google3/third_party/gemini_coder/framework/generator/generator.(*PlannerGenerator).Generate
| | third_party/gemini_coder/framework/generator/planner_generator.go:95
| | google3/third_party/gemini_coder/framework/executor/executor.(*Executor).Execute
| | third_party/gemini_coder/framework/executor/executor.go:303
| | google3/third_party/jetski/cortex/cortex.(*CascadeManager).executeHelper.func1
| | third_party/jetski/cortex/cascade_manager.go:1558
| | google3/third_party/jetski/cortex/cortex.(*CascadeManager).executeHelper.func2
| | third_party/jetski/cortex/cascade_manager.go:1676
| | runtime.goexit
| | third_party/go/gc/src/runtime/asm_amd64.s:1771
| Wraps: (3) UNAVAILABLE (code 503): No capacity available for model claude-opus-4-6-thinking on the server
| Wraps: (4) UNAVAILABLE (code 503): No capacity available for model claude-opus-4-6-thinking on the server
| Error types: (1) *go_utils.withTags (2) *withstack.withStack (3) *errutil.withPrefix (4) *utils.HTTPError
Wraps: (3) model unreachable: UNAVAILABLE (code 503): No capacity available for model claude-opus-4-6-thinking on the server: UNAVAILABLE (code 503): No capacity available for model claude-opus-4-6-thinking on the server
Wraps: (4) attached stack trace
– stack trace:
| google3/third_party/jetski/cortex/shared/shared.init
| third_party/jetski/cortex/shared/interfaces.go:21
| runtime.doInit1
| third_party/go/gc/src/runtime/proc.go:8105
| runtime.doInit
| third_party/go/gc/src/runtime/proc.go:8072
| runtime.main
| third_party/go/gc/src/runtime/proc.go:258
| runtime.goexit
| third_party/go/gc/src/runtime/asm_amd64.s:1771
Wraps: (5) model api cannot be reached
Error types: (1) *withstack.withStack (2) *secondary.withSecondaryError (3) *errutil.withPrefix (4) *withstack.withStack (5) *errutil.leafError

3 Likes

I have the same error. The worst part is when i switch back to 4.5, it says “Claude Opus 4.5 is no longer available. Please switch to Claude Opus 4.6.”. I think it P0 level bug.

1 Like

I’ve gotten that, “Claude Opus 4.5 is no longer available. Please switch to Claude Opus 4.6.” error message.

But now I am able to use Opus 4.5 again… So weird

same situation. it’s a shame that ultra user pay for nothing

I have the same error with all models… Sometimes the entire ide frozes… they are having a lot of performance issues.

1 Like

It seems that Google is setting its toys on our money, time and nerves.
I can almost see them laughing behind the monitors.

1 Like

idk what’s going on at antigravity rn. nothing works. terminal doesn’t even open sometimes. agent always terminates, due to some error.

1 Like

Problem got from bad to worse when Opus 4.6 was introduced. I was so happy with Opus 4.5, not sure if I needed the latest model.

Hi @Jonathan_Gorce,

A 503 error (UNAVAILABLE) indicates that the service is temporarily overloaded or experiencing a capacity constraint. This is common during peak usage hours, particularly with long context requests. The best immediate step is to wait some time and try again.

Could you please confirm if you are still facing this issue?

This issue still exists for me as an Ultra user, and it’s very tedious to press retry hundreds of times during small sessions.

This happens with short conversations and simple prompts.

Same for me. I’m so tired because I need to work with 3.1 pro… my Claude Opus 4.6 is not working anymore.

This problem on AI Ultra has been going on for weeks now and Google has no intention of fixing it. The Gemini 3.x is a joke compared to Opus 4.6; however, this fault makes the whole app (Antigravity) impossible to use. There is no way I will pay for another month of this product.

Sorry but this is not enough: the problem keeps raising at every hours in the day and with relatively short prompts and context windows. This very much looks like a structural problem on Google side: at present, a very premium service (AI Ultra) is unusable to me.

Hi @Fabio_Da_Soghe @AF666 @Jean-Lou_Dupont,

Thank you for bringing this to our attention. We sincerely apologize for the inconvenience this has caused. We understand that ‘retry’ workaround is not a long term solution.

To help us troubleshoot quickly, please confirm if the 503 error is specific to the Claude Opus 4.6 model or if it also occurs when using other models (such as Gemini 3.1 Pro, GPT-OSS-120b).

1 Like

@chunduriv Do you understand that we pay 200$ for nothing? It is literally waste of money. Dont say about capacity, we are top tier users, and you HAVE to reserve enough capacity for someone who is paying so much money.

Trajectory ID: 2b1e1ed2-deff-4f05-9d3b-95e4685…
Error: HTTP 503 Service Unavailable
Sherlog:
TraceID: 0x6246c8039b8d…
Headers: {“Alt-Svc”:[“h3=“:443”; ma=2592000,h3-29=“:443”; ma=2592000”],“Content-Length”:[“429”],“Content-Type”:[“text/event-stream”],“Date”:[“Tue, 10 Mar 2026 09:38:35 GMT”],“Server”:[“ESF”],“Server-Timing”:[“gfet4t7; dur=337”],“Vary”:[“Origin”,“X-Origin”,“Referer”],“X-Cloudaicompanion-Trace-Id”:[“6246c8039b8da407”],“X-Content-Type-Options”:[“nosniff”],“X-Frame-Options”:[“SAMEORIGIN”],“X-Xss-Protection”:[“0”]}

{
“error”: {
“code”: 503,
“details”: [
{
@type”: “type.googleapis.com/google.rpc.ErrorInfo”,
“domain”: “cloudcode-pa.googleapis.com”,
“metadata”: {
“model”: “claude-opus-4-6-thinking”
},
“reason”: “MODEL_CAPACITY_EXHAUSTED”
}
],
“message”: “No capacity available for model claude-opus-4-6-thinking on the server”,
“status”: “UNAVAILABLE”
}
}

I still face the issue 28 days later anyway I’ll probably cancel the subscription pretty soon this is not a working product

1 Like

My post has just been removed, where I posted about my concerns regarding the inconsistency of Antigravity and why Google isn’t providing the quality service we paid for. I am cancelling my AI Ultra subscription and leaving this forum, which is wasting my time.

1 Like

By the way, this is most prevalent to only ULTRA tier users. Free / Pro users hardly face this issue. So a fix would be to downgrade / have multiple pro accounts rather than 1 Ultra account.

Thanks for the suggestion, but downgrading simply isn’t an option for us. We are an enterprise, which means we absolutely require enterprise-level security and data protection.

There is no ‘Pro Enterprise’ tier that offers the safeguards we need. It is highly irresponsible for anyone to do business or enterprise software development without strict, enterprise-level protection for their code and IP. We paid for the Ultra tier specifically to meet those security requirements. The fact that the highest tier is the one failing—and that the only ‘fix’ is to compromise our security by downgrading—is exactly why this situation is so unacceptable.

Error: HTTP 503 Service Unavailable

“error”: {
“code”: 503,
“details”: [
{
@type”: “type.googleapis.com/google.rpc.ErrorInfo”,
“domain”: “cloudcode-pa.googleapis.com”,
“metadata”: {
“model”: “claude-opus-4-6-thinking”
},
“reason”: “MODEL_CAPACITY_EXHAUSTED”
}
],
“message”: “No capacity available for model claude-opus-4-6-thinking on the server”,
“status”: “UNAVAILABLE”
}
}

Over a month and still an issue.