GPT-4o API Rate Limits & Pricing 2026

OpenAIGenerally available

API model string

gpt-4o

Context window

128K tokens

Max output 4K–16K tokens (version-dependent)

Knowledge cutoff: October 2023
Released: 2024-05
Modalities: text, image in; text out

Last verified July 10, 2026

Rate limits by tier

OpenAI uses a five-tier spend-based system. GPT-4o is a legacy model — new integrations should use GPT-5.x. Limits below are from early-2026 third-party trackers; verify in your OpenAI dashboard as legacy model allocations may differ from current documentation.

Tier	Requirements	RPM	TPM	RPD	Concurrent	Notes
Free	No ongoing free production tier	not published	not published	not published	not published	No free production access for GPT-4o.
Tier 1	First payment (~$5 cumulative spend)	500 (per third-party trackers — confirm in your console/dashboard)	30,000 (per third-party trackers — confirm in your console/dashboard)	not published	not published	GPT-4o-class baseline. Legacy model — new projects should use GPT-5.x.
Tier 2	~$50 cumulative + 7 days account age	not published	not published	not published	not published	Automatic advancement.
Tier 3	~$100 cumulative	not published	not published	not published	not published	Automatic advancement.
Tier 4	~$250 cumulative	not published	not published	not published	not published	Automatic advancement.
Tier 5	~$1,000 cumulative + 30 days	10,000 (per third-party trackers — confirm in your console/dashboard)	30,000,000 (per third-party trackers — confirm in your console/dashboard)	not published	not published	Legacy tier ceiling; manual increase request at Tier 5 only (3–10 business-day response).
Enterprise	Contact sales	not published	not published	not published	not published	not published

Swipe the table sideways to see every limit column.

1.GPT-4o limits from early-2026 trackers (per third-party trackers — verify in your console/dashboard); official docs show GPT-5.5 as the current priority model.
2.Azure endpoint lists GPT-4o at $2.50/$10.00 — may differ from OpenAI direct; confirm in platform dashboard.
3.Rolling 60-second and 24-hour windows; no fixed reset.
4.Legacy model — continued availability not guaranteed; plan migration to GPT-5.x.

Limits verified against the OpenAI docs, July 10, 2026.

Token pricing

What you pay per million tokens (USD). Input and output are billed separately.

Input

$2.50

per 1M tokens

Output

$10.00

per 1M tokens

Batch discount

−50%

on batch jobs

Pricing based on Azure endpoint reference ($2.50/$10.00/MTok) — OpenAI's primary pricing page no longer prominently lists GPT-4o as of July 2026. Verify pricing on OpenAI direct before billing.
Cached input pricing: not published for this legacy model.
Batch API eligibility: 50% off if still eligible — verify on current OpenAI pricing page for legacy models.
No long-context surcharge published for GPT-4o (128K context only).
Data-residency uplift may apply for regional processing.

Side-project chatbot

$22.50

per month

Assumptions

5M input tokens / 1M output tokens per month (Azure reference pricing)

$2.50 × 5 + $10.00 × 1 = $12.50 + $10.00 = $22.50/month

Mid-scale app

$270

per month

Assumptions

60M input / 12M output; no cache discount confirmed for legacy model

$2.50 × 60 + $10.00 × 12 = $150 + $120 = $270/month (Azure reference; verify on OpenAI direct)

High-volume pipeline

$1,800

per month

Assumptions

400M input / 80M output per month (Azure reference)

$2.50 × 400 + $10.00 × 80 = $1,000 + $800 = $1,800/month (Azure reference; verify on OpenAI direct)

Run your own numbers

Drag your real monthly token volumes and watch the bill update live — priced against rival models at the same usage.

Input tokens / month30M

1M500M

Output tokens / month5M

0.1M100M

Estimated gpt-4o spend

$125/mo

Input: $75

Output: $50

30M in × $2.50 + 5M out × $10.00 = $125

Same volume, priced across models

gpt-4oThis model$125
Claude 3.5 Sonnet$165
GPT-5 (GPT-5.5)$300

Rivals priced at their published input/output rates for the same monthly volumes. Prompt caching is model-specific, so it is applied to gpt-4o only. Estimates for comparison; real bills vary with request shape and long-context surcharges.

gpt-4o vs the alternatives

GPT-4o (legacy) compared to its successor GPT-5.5 and the also-legacy Claude 3.5 Sonnet.

Aspect	gpt-4o	GPT-5 (GPT-5.5)	Claude 3.5 Sonnet
Input price	$2.50/MTok (Azure ref)	$5.00/MTok	~$3.00/MTok (legacy)
Output price	$10.00/MTok (Azure ref)	$30.00/MTok	~$15.00/MTok (legacy)
Context window	128K	~1M (922K)	200K
Model status	legacy	ga current	legacy
Multimodal	yes (text+image)	yes	text+image
Batch discount	50% (verify eligibility)	50%	50%
Cached input price	not published	$0.50/MTok	not published (legacy)
Knowledge cutoff	Oct 2023	Dec 2025	Apr 2024

Swipe the table sideways to see every model.

Hitting a 429? The playbook

The exact errors you'll see

429 Too Many Requests

{"error": {"message": "Rate limit reached for gpt-4o in organization org-xxx on requests per min (RPM): Limit 500, Used 500, Requested 1.", "type": "requests", "code": "rate_limit_exceeded"}}

HTTP header Retry-Afterx-ratelimit-limit-requestsx-ratelimit-remaining-requestsx-ratelimit-reset-requestsx-ratelimit-limit-tokensx-ratelimit-remaining-tokensx-ratelimit-reset-tokens

Why it happens & how to fix it

RPM burst hitting 500 req/min Tier 1 cap

Queue requests; implement sleep/delay between API calls; use exponential backoff on 429.

TPM limit hit on 128K context requests (30K TPM Tier 1 cap)

Count tokens before submission; reduce max_tokens; chunk large documents.

RPD daily cap at low tiers

Use Batch API (dedicated pool, does not touch live RPM); upgrade tier.

Multiple services sharing one API key exhausting org-wide limits

Assign separate API keys per workload; monitor per-key consumption.

Legacy model deprioritized in OpenAI's request queue

Migrate to GPT-5.x (gpt-5.5 or gpt-5.4); current-gen models receive better capacity allocation.

Retry strategy

Exponential backoff with jitter starting at 1 second, doubling up to 60 seconds. Always honor the Retry-After header value. Rolling windows mean you may only need to wait seconds, not minutes. Monitor x-ratelimit-remaining-tokens header and slow down proactively when remaining drops below 20%.

retry.ts

1import OpenAI, { RateLimitError } from "openai";
2
3const client = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });
4
5async function chatWithRetry(
6  messages: OpenAI.Chat.ChatCompletionMessageParam[],
7  maxRetries = 6
8): Promise<string> {
9  let attempt = 0;
10  while (attempt <= maxRetries) {
11    try {
12      const response = await client.chat.completions.create({
13        model: "gpt-4o", // consider migrating to "gpt-5.5"
14        messages,
15        max_tokens: 1024,
16      });
17      return response.choices[0].message.content ?? "";
18    } catch (err) {
19      if (err instanceof RateLimitError) {
20        const retryAfter = Number(err.headers?.["retry-after"] ?? 0);
21        const jitter = Math.random() * 1000;
22        const backoff = retryAfter > 0
23          ? retryAfter * 1000
24          : Math.min(1000 * 2 ** attempt, 60_000) + jitter;
25        console.warn(`429 — retrying in ${(backoff / 1000).toFixed(1)}s (attempt ${attempt + 1}/${maxRetries})`);
26        await new Promise(r => setTimeout(r, backoff));
27        attempt++;
28      } else {
29        throw err;
30      }
31    }
32  }
33  throw new Error("Max retries reached");
34}

How to raise your limits

The ladder from the starter tier to enterprise — what each rung takes, and what it unlocks.

Tier 1

Automatic

~$5 cumulative spend

Unlocks: ~500 RPM / ~30K TPM (per third-party trackers — confirm in dashboard)

Tier 2

Automatic

~$50 cumulative + 7 days account age

Unlocks: Higher RPM/TPM than Tier 1

Tier 3–4

Automatic

$100–$250 cumulative spend

Unlocks: Progressively higher limits; check dashboard for current legacy-model allocation

Tier 5

Automatic; manual limit-increase request thereafter (3–10 business-day response)

$1,000 cumulative + 30 days; then Settings → Limits → Request increase

Unlocks: ~10,000 RPM / ~30M TPM ceiling; manual raise available

Enterprise

Negotiated

Contact OpenAI sales

Unlocks: Custom throughput and SLAs

Migrate to GPT-5.x (recommended)

Immediate

Change model parameter to gpt-5.5 or gpt-5.4 — same OpenAI API shape

Unlocks: Current-gen capacity allocation + 1M context + prompt caching at $0.50/MTok + updated pricing

Cut your token spend

Migrate to GPT-5.x

1M context vs 128K; prompt caching at $0.50/MTok; current-gen capacity allocation

Change model: "gpt-4o" to model: "gpt-5.5" or model: "gpt-5.4" — the API request/response shape is identical. Verify env var for API key.

Batch API for non-real-time workloads

50% discount on input and output; dedicated pool that doesn't compete with live RPM quota

Verify GPT-4o Batch eligibility on current OpenAI pricing page. Use for classification, summarization, and data-enrichment jobs.

Reduce prompt size aggressively

30K TPM Tier-1 cap is tight — lean prompts prevent TPM exhaustion faster than RPM

Retrieve only relevant document sections; keep system prompts under 200 tokens; set max_tokens to actual needed length.

Monitor x-ratelimit-remaining-tokens proactively

Prevents 429 errors before they occur

Back off when remaining tokens drops below 20% of your tier limit rather than waiting for a 429.

Consolidate API key usage per product

Isolates quota consumption so one service can't exhaust another's limits

Assign separate API keys per product line within your org; monitor per-key consumption in the dashboard.

Use streaming (stream: true)

Reduces perceived latency without changing rate costs

Streaming gets first tokens to the user faster while the full response generates — useful for user-facing applications with tight UX requirements.

Cache common system prompts at application layer

Reduces identical API calls (GPT-4o has no confirmed prompt-cache discount)

Store and reuse responses to identical system prompt + short query combinations where freshness is not required.

Frequently asked questions

Is GPT-4o still available to use?

Yes, GPT-4o remains callable as a legacy model as of July 2026. However, it was dropped from OpenAI's current-models pricing view and is no longer recommended for new integrations. Use GPT-5.5 (gpt-5.5) for new projects.

What are the GPT-4o API rate limits?

Based on early-2026 third-party trackers, Tier 1 is approximately 500 RPM and 30,000 TPM; Tier 5 reaches roughly 10,000 RPM and 30M TPM. These figures are from community trackers — verify your actual limits in the OpenAI dashboard, as legacy model allocations may differ.

What is the GPT-4o API price per token?

Azure references GPT-4o at $2.50/MTok input and $10.00/MTok output. OpenAI's primary pricing page no longer prominently lists GPT-4o as of July 2026. Verify current pricing on OpenAI direct before billing.

How do I fix GPT-4o 429 rate limit errors?

Honor the Retry-After header on every 429 response. Implement exponential backoff with jitter starting at 1 second. The most durable fix is migrating to GPT-5.5, which receives current-gen capacity allocation and may have better rate limits for your tier.

GPT-4o vs GPT-5 pricing: is GPT-4o still cheaper?

GPT-4o's Azure reference pricing ($2.50/$10.00/MTok) is cheaper per token than GPT-5.5 ($5.00/$30.00/MTok). However, GPT-5.5 adds prompt caching at $0.50/MTok and 1M context, which can offset the per-token cost gap for cache-heavy workloads.

Will GPT-4o be shut down?

No official shutdown date has been announced for the GPT-4o API endpoint as of July 10, 2026. It is listed as a legacy model. OpenAI may add a sunset date in the future — build a model-string abstraction so you can swap to GPT-5.x in one config change.

How do I increase my GPT-4o rate limits?

Limits advance automatically at $50 (Tier 2), $100 (Tier 3), $250 (Tier 4), and $1,000 + 30 days (Tier 5). At Tier 5, submit a manual limit-increase request via Settings → Limits in your OpenAI dashboard (3–10 business-day response). The most effective increase is migrating to GPT-5.5, which receives higher current-gen capacity.

Can RapidDev help migrate our GPT-4o integration to GPT-5?

Yes. RapidDev engineers handle GPT-4o to GPT-5.5 migrations including prompt caching setup, Batch API pipelines, and rate-limit optimization. Book a free scoping call at rapidevelopers.com/contact.

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

Get a free estimate

30-min call. No commitment.

GPT-4o API Rate Limits, Pricing & Performance (July 2026)

Rate limits by tier

Token pricing

Run your own numbers

gpt-4o vs the alternatives

Hitting a 429? The playbook

How to raise your limits

Tier 1

Tier 2

Tier 3–4

Tier 5

Enterprise

Migrate to GPT-5.x (recommended)

Cut your token spend

Migrate to GPT-5.x

Batch API for non-real-time workloads

Reduce prompt size aggressively

Monitor x-ratelimit-remaining-tokens proactively

Consolidate API key usage per product

Use streaming (stream: true)

Cache common system prompts at application layer

Frequently asked questions

We build AI apps that don't hit rate limits

Your next step

Still weighing your options?

We put the rapid in RapidDev

GPT-4o API Rate Limits, Pricing & Performance (July 2026)

Rate limits by tier

Token pricing

Run your own numbers

gpt-4o vs the alternatives

Hitting a 429? The playbook

How to raise your limits

Tier 1

Tier 2

Tier 3–4

Tier 5

Enterprise

Migrate to GPT-5.x (recommended)

Cut your token spend

Migrate to GPT-5.x

Batch API for non-real-time workloads

Reduce prompt size aggressively

Monitor x-ratelimit-remaining-tokens proactively

Consolidate API key usage per product

Use streaming (stream: true)

Cache common system prompts at application layer

Related models

GPT-5 API Rate Limits & Pricing

GPT-4 Turbo API Rate Limits & Pricing

GPT-3.5 Turbo API Rate Limits & Pricing

Claude 3.5 Sonnet API Rate Limits & Pricing

More gpt-4o resources

GPT-4o 429 & 500 Error Fix Pages

GPT-4o in n8n & Zapier — Migration Guides

GPT-4o to GPT-5 Migration Implementation

Frequently asked questions

We build AI apps that don't hit rate limits

Your next step

Still weighing your options?

We put the rapid in RapidDev