API model string
gpt-4-turboContext window
128K tokens
Max output 4K tokens
- Knowledge cutoff
- April 2024
- Released
- 2023-11
- Modalities
- text, image in; text out
Last verified July 10, 2026
Rate limits by tier
GPT-4 Turbo is a delisted legacy model — no official per-tier limits are published as of July 10, 2026. Historical dashboard limits (GPT-4-class Tier 1) were approximately 500 RPM / 10,000–30,000 TPM; these may no longer reflect current allocation. All figures must be verified in your OpenAI dashboard.
| Tier | Requirements | RPM | TPM | RPD | Concurrent | Notes |
|---|---|---|---|---|---|---|
| Free | No free production tier | not published | not published | not published | not published | No free production access; model is legacy/delisted. |
| Tier 1 | ~$5 cumulative spend | not published (legacy; verify in dashboard) | not published | not published | not published | Historical GPT-4-class Tier 1 was ~500 RPM / 10K–30K TPM (pre-2026, third-party — verify); model-specific legacy limits may differ. |
| Tier 2–4 | $50–$250 cumulative spend | not published | not published | not published | not published | Verify in dashboard; limits may be lower than current-gen models. |
| Tier 5 | ~$1,000 cumulative + 30 days | not published | not published | not published | not published | Manual limit-increase request via Settings → Limits → Request increase; 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-4 Turbo is delisted from OpenAI's current pricing table — no official per-tier limits are published as of July 10, 2026.
- 2.Historical dashboard limits (pre-2026) were GPT-4-class: Tier 1 ~500 RPM / 10–30K TPM; these may no longer reflect current allocation.
- 3.OpenAI's recommended replacement is GPT-5.4, GPT-5.5, or GPT-4.1 (1M context).
- 4.Self-serve fine-tuning for GPT-4 Turbo blocked for new orgs as of May 7, 2026; existing customers until January 6, 2027.
- 5.GPT-4 Turbo pricing is not published (delisted) as of July 10, 2026.
- 6.Historical third-party data (pre-delisting, not current first-party): approximately $10.00 input / $30.00 output per MTok — verify, not current pricing.
- 7.Batch API eligibility for this legacy model: not confirmed; verify on current OpenAI pricing page.
- 8.Migration recommendation: GPT-5.5 ($5.00/$30.00/MTok with $0.50/MTok caching) or GPT-4.1 nano ($0.10/$0.40/MTok) — both cheaper than historical GPT-4 Turbo pricing.
Limits verified against the OpenAI docs, July 10, 2026.
gpt-4-turbo vs the alternatives
GPT-4 Turbo (delisted) compared to its live successors GPT-5.5 and legacy GPT-4o.
| Aspect | gpt-4-turbo | GPT-5 (GPT-5.5) | GPT-4o |
|---|---|---|---|
| Input price | not published (delisted; hist ~$10.00) | $5.00/MTok | $2.50/MTok (Azure ref) |
| Output price | not published (delisted; hist ~$30.00) | $30.00/MTok | $10.00/MTok (Azure ref) |
| Context window | 128K | ~1M (922K) | 128K |
| Model status | legacy/delisted | ga current | legacy |
| Knowledge cutoff | Apr 2024 | Dec 2025 | Oct 2023 |
| Pricing availability | not published | published | Azure reference |
| Fine-tuning | blocked for new orgs (May 7, 2026) | being wound down | being wound down |
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-4-turbo in organization org-xxx on requests per min (RPM): ...", "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-tokensWhy it happens & how to fix it
RPM exceeded — GPT-4-class historically had lower starting limits than GPT-3.5 Turbo
Exponential backoff; check exact limit in your dashboard — legacy limits may not match current published tables.
TPM exceeded on 128K context prompts
Reduce prompt length; chunk documents; set conservative max_tokens (GPT-4 Turbo's 4K max output means input management is the primary lever).
Legacy model deprioritized in OpenAI's request queue
Migrate to GPT-5.5 (gpt-5.5) or GPT-4.1 (gpt-4.1) — current-gen models receive higher capacity allocation.
Fine-tuned GPT-4 Turbo jobs no longer startable for new orgs (May 7, 2026)
Migrate fine-tuning to a supported model before the January 6, 2027 deadline for existing orgs.
Organization-shared limits exhausted by multiple services on one API key
Isolate GPT-4 Turbo traffic to a separate API key; monitor per-key usage in the dashboard.
Retry strategy
Honor the Retry-After header on every 429 response. Implement exponential backoff with jitter starting at 1 second, doubling each retry up to 60 seconds. OpenAI uses rolling 60-second windows — not fixed clock resets — so waits are typically short. Plan migration to GPT-5.5 as the long-term resolution.
1import OpenAI, { RateLimitError } from "openai";23const client = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });45async function chatWithRetry(6 messages: OpenAI.Chat.ChatCompletionMessageParam[],7 maxRetries = 68): Promise<string> {9 let attempt = 0;10 while (attempt <= maxRetries) {11 try {12 const response = await client.chat.completions.create({13 model: "gpt-4-turbo", // migrate to "gpt-5.5" or "gpt-4.1" — same API shape14 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 > 023 ? retryAfter * 100024 : 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.
Legacy Tier 1
Automatic~$5 cumulative spend
Unlocks: Historical GPT-4-class limits (dashboard is authoritative — legacy limits may differ from current published tables)
Tier 2–4
Automatic$50–$250 cumulative spend
Unlocks: Higher limits; verify in dashboard as legacy limits may not scale identically to current models
Tier 5
Automatic; manual limit-increase request thereafter (3–10 business-day response)$1,000 cumulative + 30 days; then Settings → Limits → Request increase
Unlocks: Manual limit-increase form available at Tier 5
Migrate to GPT-5.5 (recommended)
ImmediateChange model parameter to gpt-5.5 — same OpenAI API shape
Unlocks: Current-gen capacity + 1M context + prompt caching at $0.50/MTok + published pricing
Migrate to GPT-4.1
ImmediateChange model parameter to gpt-4.1
Unlocks: 1M context window with current-gen support and likely lower input price
Cut your token spend
Migrate to GPT-5.5 or GPT-4.1
GPT-4.1 offers 1M context; GPT-5.5 adds $0.50/MTok prompt caching and Dec 2025 knowledge cutoffChange model: "gpt-4-turbo" to model: "gpt-5.5" or model: "gpt-4.1" — the API request/response shape is identical. Migration is the single highest-impact optimization available.
Set max_tokens explicitly
GPT-4 Turbo's 4K max output means you rarely hit output caps; input token management is the primary leverSet max_tokens to your actual needed output length; avoid leaving it at default which may over-reserve TPM quota.
Chunk large documents
Keeps individual requests under per-request TPM estimateSplit 128K-context inputs into smaller chunks rather than sending a single maximum-context request; reduces TPM pressure per call.
Use Batch API if eligible
50% off input and output; dedicated pool not competing with live RPM quotaVerify GPT-4 Turbo Batch eligibility on the current OpenAI pricing page — legacy models may be excluded. Use for offline classification or summarization.
Monitor fine-tuning migration timeline
Avoid Jan 6, 2027 hard deadline for existing org fine-tuned GPT-4 Turbo modelsAudit all fine-tuned GPT-4 Turbo checkpoints; plan migration to a supported model before the deadline.
Consolidate API calls within 128K context
Reduces total request count, cutting RPM pressureGPT-4 Turbo has strong instruction-following; consolidate what previously required multiple calls into one larger prompt within the 128K limit.
Frequently asked questions
Is GPT-4 Turbo still available?
The GPT-4 Turbo API endpoint (gpt-4-turbo) remains callable as of July 2026, but it is delisted from OpenAI's current pricing table. No official shutdown date has been announced for API access. Fine-tuning is blocked for new organizations since May 7, 2026.
What are the GPT-4 Turbo API rate limits?
OpenAI does not publish per-tier rate limits for GPT-4 Turbo as of July 10, 2026 (delisted model). Historical GPT-4-class Tier 1 was approximately 500 RPM / 10,000–30,000 TPM — but these figures are from pre-2026 third-party trackers and may no longer reflect current allocation. Check your actual limits in the OpenAI dashboard.
What is the GPT-4 Turbo price per token?
GPT-4 Turbo pricing is not published (delisted) as of July 2026. Historical third-party data (pre-delisting) showed approximately $10.00 input / $30.00 output per MTok — but these are not current first-party figures. OpenAI recommends GPT-5.5 ($5.00/$30.00/MTok) or GPT-4.1 nano ($0.10/$0.40/MTok) as replacements.
What should I migrate GPT-4 Turbo to?
For most chat/completion use cases, migrate to GPT-5.5 (model: "gpt-5.5") — same API shape, 1M context, and published pricing. For budget-sensitive or high-volume workloads, GPT-4.1 nano (model: "gpt-4.1-nano") at $0.10/$0.40/MTok is OpenAI's recommended low-cost path.
When does GPT-4 Turbo fine-tuning end?
Self-serve fine-tuning for GPT-4 Turbo was blocked for new organizations as of May 7, 2026. Existing organizations have until January 6, 2027 to complete fine-tuning. No deadline has been announced for the base API endpoint.
How do I fix a 429 error on GPT-4 Turbo?
Check the Retry-After header on the 429 response and wait that many seconds. Implement exponential backoff with jitter (1s start, double each retry, max 60s). OpenAI uses rolling 60-second windows, so waits are usually short. Long-term: migrate to GPT-5.5 for better capacity allocation.
GPT-4 Turbo vs GPT-5: which should I use?
GPT-5.5 for all new projects. GPT-4 Turbo is delisted with no published pricing, has a shorter 128K context, older knowledge cutoff (April 2024), and receives lower capacity allocation as a legacy model. GPT-5.5 is cheaper per input token than historical GPT-4 Turbo pricing and offers 1M context.
Can RapidDev help migrate our GPT-4 Turbo integration?
Yes. RapidDev engineers handle GPT-4 Turbo to GPT-5.5 migrations including API updates, prompt caching setup, and fine-tuning migration planning. 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
30-min call. No commitment.