Is Antigravity CLI down or slower today?

As of 2026-11-11 13:38 UTC · an hourly probe of Antigravity CLI (Gemini 3.8 Flash) and Antigravity CLI (GPT-OSS 120B) on the Google AI subscription, one account, from one EU server · not an official status page

Antigravity CLI on the Google AI subscription: answering

Last hourly probe — Gemini 3.8 Flash: answered in 98.4 s at 13:38 UTC · GPT-OSS 120B: answered in 17.4 s at 13:37 UTC

Is Antigravity CLI answering on the Google AI subscription right now? With Gemini 3.8 Flash: yes — our last probe, at 13:38 UTC, answered in 98.4 s; 0 of 24 calls failed in the last 24 hours. With GPT-OSS 120B: yes — our last probe, at 13:37 UTC, answered in 17.4 s; 0 of 2 calls failed in the last 24 hours. This is an hourly probe of one account from one EU server, not a status page: it can miss an outage between probes and does not see problems that hit other accounts or regions. Is Antigravity CLI slower today? Antigravity CLI (Gemini 3.8 Flash) generated 152.3 tokens/s over the last 24 hours vs 158.0 tokens/s in the 7 days before (−4%) — no significant change (p=0.0303). Antigravity CLI (GPT-OSS 120B): measuring, 2 samples in the last 24 hours; no verdict until the 7-day baseline is in.

Also: Codex vs Claude Code, side by side · all coding agents · Is Claude Code down or slower today? · Is Codex down or slower today?

Is Antigravity CLI slower today?

Each model is compared only with its own last 7 days on one fixed task with a single exact answer.

Antigravity CLI · Gemini 3.8 Flash: no significant change in the last 24 hours

Last measured 2026-11-11 13:38 UTC · agy 1.5.5 · effort high · Google AI subscription · errors in 24 h: 0% · requested gemini-3.8-flash-high; the CLI does not report which model answered, so a silent substitution would not be visible · thinking is not streamed by the CLI, so the first token is the first visible token after thinking; tools are denied by permission policy, a tool attempt counts as an error; the hard panel is asked in one slice a day (15-16 questions after 00:47 UTC, the whole panel every 5 days) because Flash spends 5-45k tokens per question

Metric	Last 24 h	7 days before	Change	Verdict
Speed	152.3 tokens/s	158.0	−4%	no significant change (p=0.0303)
Thinking	13,594 tokens	14,773	−8%	no significant change (p=0.0659)
First token	89.6 s	93.3 s	−4%	no significant change (p=0.359)

Antigravity CLI · GPT-OSS 120B: measuring — the 7-day baseline is still filling up

Last measured 2026-11-11 13:37 UTC · agy 1.5.5 · effort medium · Google AI subscription · errors in 24 h: 0% · requested gpt-oss-120b-medium; the CLI does not report which model answered, so a silent substitution would not be visible · an open-weight model served by Google inside Antigravity CLI on the same plan as Gemini 3.8 Flash; the CLI reports no thinking tokens for it (it does reason: the first visible token comes after the reasoning), so thinking is not measured; tools are denied by permission policy; the hard panel is asked in one slice a day (39 questions after 00:47 UTC, the whole panel every 2 days)

Metric	Last 24 h	7 days before	Change	Verdict
Speed	— tokens/s	—	—	measuring, no verdict yet
Thinking	— tokens	—	—	not reported by the CLI
First token	—	—	—	measuring, no verdict yet

Speed is output tokens (thinking included) per second of response time. A change counts as significant only when the median moves by at least 15% and a Mann-Whitney test gives p < 0.01, with at least 12 recent and 48 baseline samples. These numbers are not comparable with other agents: Claude Code reports API time while Codex and Antigravity CLI report turn time, Antigravity CLI does not stream thinking, and the tokenizers differ, so the same number means different things.

Hourly speed, last 7 days

Antigravity CLI · Gemini 3.8 Flash

Antigravity CLI · GPT-OSS 120B

Daily medians

UTC day	Gemini 3.8 Flash tokens/s	thinking	GPT-OSS 120B tokens/s	thinking
2026-11-11	151.6	13,173	163.9	—
2026-11-10	153.0	14,212	—	—
2026-11-09	155.4	15,388	—	—
2026-11-08	165.0	14,472	—	—
2026-11-07	161.3	15,466	—	—
2026-11-06	158.9	14,068	—	—
2026-11-05	154.3	15,462	—	—
2026-11-04	151.7	14,763	—	—

How long does Antigravity CLI take to answer?

Model	Median, 24 h	Median, 7 days	Failed, 24 h	Failed, 7 days
Gemini 3.8 Flash	100.6 s 24 runs	102.8 s 167 runs	0 of 24	1 of 168
GPT-OSS 120B	16.9 s 2 runs, too few	16.9 s 2 runs, too few	0 of 2	0 of 2

End to end on our fixed task: from launching the CLI to the complete answer, by our own clock. The same task in the other agents: Codex vs Claude Code, side by side.

"Too few": not enough successful runs yet to compare with other agents. "Failed": calls that errored, timed out, were rate-limited or could not count as a clean measurement.

Has it got worse at hard questions?

Antigravity CLI (Gemini 3.8 Flash): 66/123 correct (54%) so far; the baseline is still being collected (day 7 of 10), so no verdict until 2026-11-14.
Antigravity CLI (GPT-OSS 120B): 17/38 correct (45%) so far; the baseline is still being collected (day 1 of 10), so no verdict until 2026-11-21.

A fixed panel of hard questions, judged against each model's own baseline; details on all coding agents.

Is this the official status?

No. This is an hourly probe of one account from one EU server, not a status page: it can miss an outage between probes and does not see problems that hit other accounts or regions. The vendor's developer API, measured separately every 5 minutes from 4 regions: Is the Gemini API (Google) down? — access through the Google AI subscription is a different service with its own limits, so the two can differ.

Questions

Is Antigravity CLI down right now?

With Gemini 3.8 Flash: not on our account — our last probe, at 13:38 UTC, answered in 98.4 s; 0 of 24 calls failed in the last 24 hours. With GPT-OSS 120B: not on our account — our last probe, at 13:37 UTC, answered in 17.4 s; 0 of 2 calls failed in the last 24 hours. This is an hourly probe of one account from one EU server, not a status page: it can miss an outage between probes and does not see problems that hit other accounts or regions.

Is Antigravity CLI slower today?

Antigravity CLI (Gemini 3.8 Flash) generated 152.3 tokens/s over the last 24 hours vs 158.0 tokens/s in the 7 days before (−4%) — no significant change (p=0.0303). Antigravity CLI (GPT-OSS 120B): measuring, 2 samples in the last 24 hours; no verdict until the 7-day baseline is in.

How long does Antigravity CLI take to answer?

On our fixed task, end to end, Antigravity CLI took a median of 100.6 s (24 runs) over the last 24 hours and 102.8 s (167 runs) over 7 days with Gemini 3.8 Flash; 16.9 s (2 runs, too few to compare) over the last 24 hours and 16.9 s (2 runs, too few to compare) over 7 days with GPT-OSS 120B. The same task in the other agents: Codex vs Claude Code, side by side.

Has Gemini 3.8 Flash in Antigravity CLI got worse at hard questions?

Antigravity CLI (Gemini 3.8 Flash): 66/123 correct (54%) so far; the baseline is still being collected (day 7 of 10), so no verdict until 2026-11-14.

Has GPT-OSS 120B in Antigravity CLI got worse at hard questions?

Antigravity CLI (GPT-OSS 120B): 17/38 correct (45%) so far; the baseline is still being collected (day 1 of 10), so no verdict until 2026-11-21.

Which model and plan does this page measure?

Antigravity CLI with Gemini 3.8 Flash (effort high); Antigravity CLI with GPT-OSS 120B (effort medium), on the Google AI subscription, CLI agy 1.5.5; one account, one EU server.

How it is measured

What: Antigravity CLI with Gemini 3.8 Flash (effort high); Antigravity CLI with GPT-OSS 120B (effort medium); tools disabled or denied by permission policy, a fresh isolated session per call, on the Google AI subscription.
When: every hour at :37 UTC, one fixed task per agent.
Where: one server in the EU (Hetzner, Germany); a single account per vendor.
Errors: a failed call is 'not measured', never a wrong answer.
Data: /api/coding-agents.json (JSON, CC BY 4.0) · markdown