API model string
imagen-4.0-generate-001Context window
n/a (image prompt)
Max output n/a
- Knowledge cutoff
- n/a
- Released
- 2024 (GA)
- Modalities
- image generation (text-to-image)
Last verified July 10, 2026
Rate limits by tier
Imagen 4 enforces per-image quotas per project in AI Studio. Google does not publish stable RPM or RPD numbers — active limits are project-specific and visible in your AI Studio console. The model is shutting down August 17, 2026.
| Tier | Requirements | RPM | TPM | RPD | Concurrent | Notes |
|---|---|---|---|---|---|---|
| Free (AI Studio) | Google account; no billing required | not published | n/a (image model) | not published | not published | Image-generation quotas exist per project in AI Studio; exact limits not published in Google docs. Not recommended for new integrations — sunset Aug 17, 2026. |
| Paid (PAYG Developer API) | Google Cloud billing account linked to project | not published | n/a (image model) | not published | not published | Per-image billing: $0.02 Fast / $0.04 Standard / $0.06 Ultra. Project-specific quotas shown in AI Studio console. Contact Google Cloud for higher limits. Sunset Aug 17, 2026 — no new integrations. |
| Enterprise | Google Cloud Sales agreement | negotiated | n/a | negotiated | negotiated | Dedicated capacity and committed use discounts. Not advisable given Aug 17, 2026 shutdown. |
Swipe the table sideways to see every limit column.
- 1.Imagen 4 is a per-image billed service, not token-billed — input/output token prices do not apply.
- 2.Quotas are enforced per project in AI Studio. Actual limits vary by project and are not publicly documented.
- 3.Imagen 4 shuts down August 17, 2026. No new project integrations are recommended. Migrate to Gemini 2.5 Flash Image.
- 4.Imagen 4 uses per-image billing, not per-token. Fast tier: $0.02/image. Standard tier: $0.04/image. Ultra tier: $0.06/image.
- 5.No batch discount published for Imagen 4.
- 6.Model shuts down August 17, 2026 — budget accordingly and plan migration costs.
Limits verified against the Google docs, July 10, 2026.
imagen-4.0-generate-001 vs the alternatives
Imagen 4 compared against its successor and the leading alternative — both with active development beyond August 2026.
| Aspect | imagen-4.0-generate-001 | Gemini 2.5 Flash Image | OpenAI GPT Image 1 |
|---|---|---|---|
| Price per image (standard) | $0.04 | not published (Gemini 2.5 Flash Image) | $0.134/1K high-res (GPT Image 1) |
| Price per image (fast/entry tier) | $0.02 | not published | ~$0.040 (GPT Image 1 low) |
| API availability after Aug 2026 | shut down Aug 17, 2026 | GA — recommended migration path | GA |
| Image quality | GA-grade (before sunset) | next-gen, announced as improved | high (GPT Image 1) |
| Batch discount | not published | 50% on Gemini models | not published |
| Active development | none (shutting down) | active | active |
Swipe the table sideways to see every model.
Hitting a 429? The playbook
The exact errors you'll see
429 Too Many RequestsQuota exceeded for quota metric 'generate_requests_per_minute'RESOURCE_EXHAUSTEDWhy it happens & how to fix it
Project RPM quota exceeded
Reduce request frequency or request a quota increase in Google Cloud Console → IAM & Admin → Quotas.
Daily image quota hit on free tier
Switch to a paid billing account or wait for quota reset. Note that the model shuts down Aug 17, 2026.
API key scope restrictions
Ensure the API key has Generative Language API enabled in Google Cloud Console → APIs & Services.
Retry strategy
Google AI documentation explicitly recommends exponential backoff. Start at 1 second, double on each retry up to a maximum of 60 seconds. Add ±20% jitter to avoid thundering-herd. Honor the Retry-After header when present. Given the August 17, 2026 shutdown, prioritize migrating to the successor rather than building retry infrastructure for Imagen 4.
1// retry.ts — Imagen 4 image generation with exponential backoff2// Base URL: https://generativelanguage.googleapis.com/v1beta/models/imagen-4.0-generate-001:predict3// NOTE: Imagen 4 shuts down August 17, 2026. Migrate to successor before then.45const API_KEY = process.env.GOOGLE_API_KEY!;6const BASE_URL =7 'https://generativelanguage.googleapis.com/v1beta/models/imagen-4.0-generate-001:predict';89async function generateImage(prompt: string, maxRetries = 5): Promise<unknown> {10 let delay = 1000;11 for (let attempt = 0; attempt <= maxRetries; attempt++) {12 const response = await fetch(`${BASE_URL}?key=${API_KEY}`, {13 method: 'POST',14 headers: { 'Content-Type': 'application/json' },15 body: JSON.stringify({ instances: [{ prompt }], parameters: { sampleCount: 1 } }),16 });17 if (response.ok) return response.json();18 if (response.status !== 429 || attempt === maxRetries) {19 throw new Error(`Imagen 4 error ${response.status}: ${await response.text()}`);20 }21 const retryAfter = response.headers.get('Retry-After');22 const waitMs = retryAfter ? parseInt(retryAfter) * 1000 : delay * (0.8 + Math.random() * 0.4);23 await new Promise((r) => setTimeout(r, waitMs));24 delay = Math.min(delay * 2, 60_000);25 }26 throw new Error('Max retries exceeded');27}How to raise your limits
The ladder from the starter tier to enterprise — what each rung takes, and what it unlocks.
AI Studio Free
ImmediateDefault quota visible in AI Studio per project; no card needed
Unlocks: Basic image generation up to project quota
PAYG Paid
Immediate after billing addedAdd a Google Cloud billing account to your project in AI Studio
Unlocks: Higher per-project quotas and SLA; per-image billing at $0.02–$0.06
Higher limits (before shutdown)
1–5 business daysGoogle Cloud Console → IAM & Admin → Quotas → request increase
Unlocks: Custom RPM/RPD caps — only worthwhile if migration cannot happen before Aug 17, 2026
Migrate to successor (mandatory)
Must complete before Aug 17, 2026Switch API string to Gemini 2.5 Flash Image before August 17, 2026
Unlocks: Continuity of image generation service; next-generation quality
Cut your token spend
Migrate immediately
Avoid 100% production outage on Aug 17, 2026Switch integration to Gemini 2.5 Flash Image (successor). The migration is the single most important action for any team using Imagen 4.
Use Fast tier for drafts
3× cost reduction vs Ultra ($0.02 vs $0.06/image)Route draft/preview generation to imagen-4.0-fast-generate-001 and only use Standard/Ultra for final production images.
Cache generated images in Cloud Storage
Eliminates regeneration cost entirelyStore generated images in Google Cloud Storage or similar. Retrieval cost is near-zero; regeneration always incurs image quota and billing.
Optimize prompts to maximize hit rate
Fewer retries = lower cost and quota consumptionTest prompt templates to achieve the desired output on the first generation. Each retry costs the same as a fresh request.
Monitor quota usage actively
Avoid unexpected quota blocksCheck AI Studio dashboard and set billing alerts in Google Cloud Console. With shutdown imminent, any quota issue could disrupt final migration testing.
Batch generation before shutdown
Pre-generate static assets before Aug 17, 2026If you have known image assets to generate, do so in bulk now and store them. After sunset, the API will be unavailable.
Frequently asked questions
Is Imagen 4 still available as an API?
Yes, but only until August 17, 2026. After that date, the Imagen 4 API will be fully shut down. Do not start new integrations. If you have existing code using imagen-4.0-generate-001, migrate to the successor model before the deadline.
What replaces Imagen 4 after the August 17, 2026 shutdown?
Google's announced successor is Gemini 2.5 Flash Image (internally known as Nano Banana). It is the recommended migration path for teams currently on Imagen 4. Check Google's official documentation for the exact API string at migration time.
How much does Imagen 4 cost per image?
Imagen 4 uses per-image billing (not token-based): Fast tier is $0.02/image, Standard is $0.04/image, and Ultra is $0.06/image, as verified July 10, 2026. There is no published batch discount.
What are the rate limits for Imagen 4?
Google does not publish stable RPM or RPD numbers for Imagen 4. Your active quota is project-specific and visible in the AI Studio console. Contact Google Cloud Support to request a quota increase.
What does the 'RESOURCE_EXHAUSTED' error mean for Imagen 4?
RESOURCE_EXHAUSTED means your project has hit its quota for image generation requests. To fix it: (1) reduce request frequency, (2) request a quota increase in Google Cloud Console → IAM & Admin → Quotas, or (3) add exponential backoff to your retry logic.
How do I migrate from Imagen 4 before the shutdown?
Update your API endpoint and model ID to the successor (Gemini 2.5 Flash Image). Test the new integration in a non-production environment before the August 17, 2026 cutoff. Teams needing help scoping the migration can reach out to RapidDev at rapidevelopers.com/contact.
Can I still use Imagen 4 after August 17, 2026?
No. August 17, 2026 is the hard shutdown date. After that date, all API calls to Imagen 4 endpoints will fail. There is no grace period — migrate before the deadline.
We build AI apps that don't hit rate limits
- Retry, backoff & caching built in
- Multi-provider fallback routing
- Fixed price, you own the code
30-min call. No commitment.