Home / Guides / Gemini Rate Limits
Gemini Rate Limits Explained
Everything you need to know about Google Gemini rate limits - RPM, RPD, TPM caps per model, free vs paid tier quotas, and how to avoid silent 429s when working from the Gemini CLI.
Last updated October 2026 · By Soren Starck
What Are Gemini Rate Limits?
Gemini enforces limits on three axes: RPM (requests per minute), RPD (requests per day), and TPM (tokens per minute). Each model has its own per-tier caps, applied per project rather than per API key, and Google now shows your active limits in Google AI Studio instead of publishing a per-model table.
Hit any of the three and Google returns a 429 Resource Exhausted with a retry-after. The Gemini CLI usually retries quietly, but during heavy multi-turn sessions you can stack failures and end up waiting longer than necessary.
On the Gemini CLI specifically: since June 18, 2026 it no longer serves free Gemini Code Assist for individuals, or Google AI Pro and Ultra. It still works with a Gemini Code Assist Standard or Enterprise license or a paid Gemini API key, and on the API key route the limits below apply.
Free vs Tier 1 Limits (Earlier Published Figures)
These figures come from Google's earlier published rate-limit table, as recorded here in April 2026. Google no longer publishes per-model numbers: your active limits are shown in Google AI Studio and depend on your usage tier, so check there before planning around any figure below.
| Model | Free RPM / RPD | Tier 1 RPM / RPD | Context |
|---|---|---|---|
| Gemini 2.5 Pro | 5 / 100 | 150 / 1,000 | 1M tokens |
| Gemini 2.5 Flash | 10 / 250 | 1,000 / 10,000 | 1M tokens |
| Gemini 2.5 Flash-Lite | 15 / 1,000 | 4,000 / 30,000 | 1M tokens |
Numbers shift as Google rebalances. Always cross-check your limits in AI Studio. The pattern in that table: Pro tightest, Flash middle, Flash-Lite most permissive.
What Happens When You Hit the Limit
- 429 errors - “Resource exhausted” with a retry-after
- CLI silent retries - your session feels stuck without explanation
- RPD lockout - once daily quota is gone, you wait until midnight Pacific time
- No native warning - neither AI Studio nor the CLI alerts you in real time
SessionWatcher
429s break your flow.
SessionWatcher prevents them.
Native app for the Mac menu bar and the Windows taskbar. Track Claude, Codex and 20 more tools' usage, costs, and rate limits in real time.
“Fast, simple, and does exactly what it should. Definitely worth it.”
@nicojerome on GitHub
macOS 14+. Free for 7 days, no credit card.
How to Monitor Gemini Rate Limits
SessionWatcher for Gemini watches your CLI activity live and surfaces the three caps in your Mac menu bar or Windows taskbar.
- Live RPM gauge - see how close to throttle you are right now
- Daily request count against RPD cap
- Token-per-minute tracking for long-context calls
- Per-model breakdown - Pro vs Flash vs Flash-Lite
- Notifications at 70% / 90% by default so you can pace
Tips for Managing Gemini Rate Limits
- Pace RPM in long CLI sessions - Pro caps are the tightest and max out fast.
- Drop to Flash for cheap turns - save Pro for the hard reasoning steps.
- Watch TPM on long contexts - a 500K-token request is cheap on RPM but expensive on TPM.
- Plan around RPD - daily caps reset at midnight Pacific time, not your local timezone.
- Upgrade when burn rate exceeds free tier - in Google's earlier published table, Tier 1 RPM was 30x free for Pro and 100x for Flash.
Frequently Asked Questions
What are Gemini rate limits?
Three axes per model: RPM, RPD, TPM. Google now shows your per-model limits in AI Studio; in its earlier published table, free Pro was the tightest (~5 RPM / 100 RPD) and Tier 1 raised everything 30–100x. RPD resets at midnight Pacific time.
How do I get past Gemini 429 errors?
Pace below RPM cap, switch to Flash, or upgrade. SessionWatcher shows live RPM so you can self-throttle.
Does the 1M context window change rate limits?
No. RPM and RPD count requests, not tokens. TPM is the cap to watch on long contexts.