# GPT-4 Turbo explained: specs, GPT-4o comparison and the October 2026 shutdown

> GPT-4 Turbo, OpenAI's 128K-context GPT-4 from November 2023, shuts down in the API on October 23, 2026. How it compares with GPT-4o and how to migrate.

- URL: https://computese.com/openai-unveils-gpt-4-turbo-the-most-powerful-ai-yet/
- Author: Duong Quan Nguyen, CEO, Computese
- Published: 2023-12-09
- Updated: 2026-09-25
- Topics: AI & automation

## In short
- GPT-4 Turbo is the version of GPT-4 that OpenAI announced at DevDay on November 6, 2023: a 128,000-token context window, knowledge up to April 2023, JSON mode, and input tokens at a third of GPT-4's price.
- Its stable snapshot, gpt-4-turbo-2024-04-09, added image input on April 9, 2024. GPT-4o followed on May 13, 2024, at half GPT-4 Turbo's price with five times the rate limits.
- GPT-4 Turbo is deprecated: OpenAI shuts down gpt-4-turbo, gpt-4-turbo-2024-04-09 and gpt-4-0613 on October 23, 2026 and names gpt-5.6-sol as the substitute. gpt-4-turbo-preview already shut down on March 26, 2026.
- As of September 2026, OpenAI's flagship models are GPT-6 Astra, Sol and Luna: reasoning models with a 1.05 million-token context window, priced from $0.10 to $10 per million input tokens.
- Migrate by finding every model name in your code, pinning a snapshot and running an evaluation set built from your own cases against each candidate. Leaderboard ranks do not predict results on your work.

GPT-4 Turbo is the cheaper, long-context version of GPT-4 that OpenAI [announced at DevDay on November 6, 2023](https://openai.com/index/new-models-and-developer-products-announced-at-devday/): a 128,000-token context window, knowledge of events up to April 2023, and input tokens at a third of GPT-4's price. It is now deprecated, and [OpenAI shuts it down in the API](https://developers.openai.com/api/docs/deprecations) on October 23, 2026.

This guide covers what GPT-4 Turbo was, how its model names map to snapshots, how it compares with GPT-4 and GPT-4o, what OpenAI has released since, the exact shutdown dates, and how to move a production system to a newer model without trusting leaderboard headlines. Prices and model lists are as of September 25, 2026, from OpenAI's own documentation. For how models like it are built and trained in the first place, start with [what generative AI is](https://computese.com/the-rise-of-generative-ai/).

## What OpenAI announced at DevDay 2023

GPT-4 had been out since March 2023 when OpenAI launched GPT-4 Turbo as a preview at DevDay. The announcement listed the changes that mattered to developers:

- **A 128K context window**, up from the 8K and 32K versions of GPT-4. OpenAI put 128,000 tokens at more than 300 pages of text in one prompt.
- **Newer knowledge.** The model knew about world events up to April 2023.
- **Lower prices.** $0.01 per 1,000 input tokens and $0.03 per 1,000 output tokens: three times cheaper for input and twice as cheap for output as GPT-4 8K.
- **JSON mode and better function calling.** A new `response_format` parameter constrained the output to valid JSON, and the model could call several functions in one message instead of one per round trip.
- **Reproducible outputs.** A beta `seed` parameter made the model return the same completion for the same request most of the time, useful for debugging and tests. Log probabilities for output tokens were promised within weeks.

Developers called the text model as `gpt-4-1106-preview`, and OpenAI promised a stable version for production within weeks. Image input came as a separate preview, `gpt-4-vision-preview`, with vision planned for the stable release. OpenAI also doubled the tokens-per-minute rate limits for all paying GPT-4 customers.

| Model at DevDay | Context window | Input, per 1,000 tokens | Output, per 1,000 tokens |
| --------------- | -------------- | ----------------------- | ------------------------ |
| GPT-4 8K        | 8K             | $0.03                   | $0.06                    |
| GPT-4 32K       | 32K            | $0.06                   | $0.12                    |
| GPT-4 Turbo     | 128K           | $0.01                   | $0.03                    |

The same day brought the Assistants API, DALL·E 3 and text-to-speech (`tts-1`) to the API, and OpenAI said the Assistants API was built on the same capabilities as its new GPTs product. Those were separate products: text-to-speech is its own model, and GPT-4 Turbo itself reads text and images and writes text. API prices are now quoted per million tokens, so GPT-4 Turbo's launch price reads $10 input and $30 output.

## How GPT-4 Turbo's model names map to snapshots

OpenAI ships a model as dated snapshots, fixed versions whose behaviour does not change, plus aliases that point at one of them. Its [model pages](https://developers.openai.com/api/docs/models/gpt-4-turbo) present snapshots as the way to hold a model version still, so its performance and behaviour stay the same. GPT-4 Turbo went through three previews and a stable release in five months:

| Model name                                          | What it was              | Released     | Status in the OpenAI API        |
| --------------------------------------------------- | ------------------------ | ------------ | ------------------------------- |
| `gpt-4-1106-preview`                                | The DevDay preview       | Nov 6, 2023  | Listed for shutdown (see below) |
| `gpt-4-vision-preview`, `gpt-4-1106-vision-preview` | Image-input preview      | Nov 6, 2023  | Shut down Dec 6, 2024           |
| `gpt-4-0125-preview`, alias `gpt-4-turbo-preview`   | Updated preview          | Jan 25, 2024 | Shut down Mar 26, 2026          |
| `gpt-4-turbo-2024-04-09`, alias `gpt-4-turbo`       | Stable model with vision | Apr 9, 2024  | Shuts down Oct 23, 2026         |

The release dates come from [OpenAI's API changelog](https://developers.openai.com/api/docs/changelog), which records GPT-4 Turbo with Vision reaching general availability on April 9, 2024. The stable snapshot accepts text and images, writes text, has a 128,000-token context window and a knowledge cutoff of December 1, 2023, later than the preview's April 2023.

The context window is shared between what you send and what the model writes. GPT-4 Turbo could take in a long contract or a large slice of a codebase in one request, but each response stopped at 4,096 output tokens. Tasks that produce long output, such as rewriting a whole document, had to be split across several calls.

![A tall stack of document pages slides into the wide mouth of a funnel, and a single short orange page comes out of the funnel's narrow spout.](https://computese.com/images/blog/openai-unveils-gpt-4-turbo-the-most-powerful-ai-yet/context-window.8d004c0552-1536.webp)

*Wide on the way in, narrow on the way out: GPT-4 Turbo read up to 128,000 tokens but wrote at most 4,096 per response.*

The alias is the trap. When a new snapshot ships, OpenAI can move the alias to it, and every request that uses the alias gets a different model without a code change. The changelog records such moves: on January 13, 2026, for example, the `gpt-4o-mini-tts` and `gpt-4o-mini-transcribe` aliases were pointed at new December 2025 snapshots, and the previous versions stayed reachable only by their dated names. A pinned snapshot gives up automatic improvements in exchange for predictability, which is what a production system needs.

![A laptop is cabled to one of three model servers, its plug locked with an orange padlock, while an orange pin, moved along a dashed arc, now marks the newest one.](https://computese.com/images/blog/openai-unveils-gpt-4-turbo-the-most-powerful-ai-yet/snapshot-pin.1968711866-1536.webp)

*An alias follows OpenAI's releases; a pinned snapshot changes only when you change it, or when its shutdown date arrives.*

## GPT-4 Turbo vs GPT-4o

[GPT-4o](https://openai.com/index/hello-gpt-4o/) arrived on May 13, 2024, five weeks after GPT-4 Turbo's stable release. OpenAI said that in the API it was twice as fast as GPT-4 Turbo, half the price and had five times the rate limits. OpenAI's own [GPT-4 Turbo page](https://developers.openai.com/api/docs/models/gpt-4-turbo) now recommends a newer model such as GPT-4o. Side by side, from OpenAI's model pages as of September 2026:

|                               | GPT-4 Turbo              | [GPT-4o](https://developers.openai.com/api/docs/models/gpt-4o)       |
| ----------------------------- | ------------------------ | -------------------------------------------------------------------- |
| Default snapshot              | `gpt-4-turbo-2024-04-09` | `gpt-4o-2024-08-06`                                                  |
| First released                | Nov 6, 2023 (preview)    | May 13, 2024                                                         |
| Input                         | Text, image              | Text, image                                                          |
| Context window                | 128,000 tokens           | 128,000 tokens                                                       |
| Maximum output per response   | 4,096 tokens             | 16,384 tokens                                                        |
| Knowledge cutoff              | Dec 1, 2023              | Oct 1, 2023                                                          |
| Price per 1M tokens, in / out | $10 / $30                | $2.50 / $10                                                          |
| Cached input discount         | Not offered              | $1.25 per 1M tokens                                                  |
| Structured Outputs            | No, JSON mode only       | Yes                                                                  |
| API status, September 2026    | Shuts down Oct 23, 2026  | First snapshot shuts down Oct 23, 2026; later snapshots have no date |

Three differences matter in practice:

- **Price.** The first GPT-4o snapshot, `gpt-4o-2024-05-13`, still lists at $5 input and $15 output, half of GPT-4 Turbo. The current `gpt-4o` default costs a quarter of GPT-4 Turbo's input price.
- **Structured Outputs.** JSON mode only guarantees valid JSON. [Structured Outputs](https://developers.openai.com/api/docs/guides/structured-outputs), launched on August 6, 2024 and available from GPT-4o on, also guarantees that the JSON matches the schema you supply, and OpenAI recommends it over JSON mode whenever a model supports it. For an integration that parses model output into a database or an API call, that is the upgrade that removes a whole class of errors.
- **Knowledge.** Newer does not mean more recent. GPT-4o's knowledge cutoff is two months earlier than the stable GPT-4 Turbo's.

The comparison with GPT-4 itself is simpler. OpenAI's [GPT-4 page](https://developers.openai.com/api/docs/models/gpt-4) lists `gpt-4-0613` with an 8,192-token context window at $30 input and $60 output per million tokens; GPT-4 Turbo has about 16 times the context at a third of the input price.

## What OpenAI released after GPT-4 Turbo

GPT-4o replaced GPT-4 Turbo six months after DevDay, and new models have followed every few months since, according to the changelog and OpenAI's announcements:

| Date                 | Release                      | What changed for developers                                   |
| -------------------- | ---------------------------- | ------------------------------------------------------------- |
| Nov 6, 2023          | GPT-4 Turbo preview          | 128K context, JSON mode, lower prices                         |
| Apr 9, 2024          | GPT-4 Turbo with Vision (GA) | Stable snapshot with image input                              |
| May 13, 2024         | GPT-4o                       | Half GPT-4 Turbo's price, faster, higher rate limits          |
| Sep 12, 2024         | o1-preview and o1-mini       | The first o-series reasoning models                           |
| Jan 31, 2025         | o3-mini                      | A small reasoning model for science, math and coding          |
| Feb 27, 2025         | GPT-4.5 preview              | A large research preview, switched off on July 14, 2025       |
| Apr 14, 2025         | GPT-4.1                      | Up to 1 million tokens of context, June 2024 knowledge        |
| Apr 16, 2025         | o3 and o4-mini               | Reasoning models aimed at math, science and coding            |
| Aug 7, 2025          | GPT-5                        | Reasoning with adjustable effort, 400,000-token context       |
| Nov 2025 to Apr 2026 | GPT-5.1, 5.2, 5.4 and 5.5    | Four more GPT-5.x flagship releases                           |
| Jul 9, 2026          | GPT-5.6 Sol, Terra and Luna  | Three tiers; Sol is OpenAI's named substitute for GPT-4 Turbo |
| Sep 3 and 22, 2026   | GPT-6 Astra, Sol and Luna    | The current flagships, 1.05 million-token context             |

[OpenAI introduced o1](https://openai.com/index/introducing-openai-o1-preview/) on September 12, 2024 as a new series of models that reason at length before they answer. [GPT-4.1](https://openai.com/index/gpt-4-1/) then stretched the context window: up to 1 million tokens, a June 2024 knowledge cutoff and, OpenAI said, 26% lower cost than GPT-4o for median queries. With [GPT-5](https://developers.openai.com/api/docs/models/gpt-5) the GPT and reasoning lines merged: it is a reasoning model with an adjustable effort setting and a 400,000-token context window.

A [reasoning model](https://developers.openai.com/api/docs/guides/reasoning) generates reasoning tokens before its answer. You do not see them, but they occupy space in the context window and are billed as output tokens, and OpenAI recommends reserving at least 25,000 tokens for reasoning and output when you start. That changes cost planning more than any price cut: depending on the problem, the reasoning behind one answer can run from a few hundred to tens of thousands of tokens. How reasoning models reach an answer, and where they still fail, is covered in our post on [logical AI](https://computese.com/exploring-the-new-frontier-in-ai-logical/).

The o-series has since been folded back in: OpenAI's [o3 page](https://developers.openai.com/api/docs/models/o3) names GPT-5 as its successor, and the flagship models on its [models page](https://developers.openai.com/api/docs/models) as of September 2026 are GPT-6 Astra, GPT-6 Sol and GPT-6 Luna.

## Is GPT-4 Turbo deprecated? Shutdown dates

Yes. On [OpenAI's deprecations page](https://developers.openai.com/api/docs/deprecations), a notice dated April 22, 2026 lists `gpt-4-turbo`, its snapshot `gpt-4-turbo-2024-04-09` and `gpt-4-turbo-completions` for shutdown on October 23, 2026. The same notice retires the original GPT-4 and the first GPT-4o snapshot on the same day. The full picture for the GPT-4 generation:

| Model or snapshot                                   | Shutdown date                                   | OpenAI's substitute       |
| --------------------------------------------------- | ----------------------------------------------- | ------------------------- |
| `gpt-4-vision-preview`, `gpt-4-1106-vision-preview` | Dec 6, 2024                                     | `gpt-4o`                  |
| `gpt-4-32k`, `gpt-4-32k-0613`, `gpt-4-32k-0314`     | Jun 6, 2025                                     | `gpt-4o`                  |
| `gpt-4-0314`                                        | Mar 26, 2026                                    | `gpt-5` or `gpt-4.1`      |
| `gpt-4-0125-preview` (`gpt-4-turbo-preview`)        | Mar 26, 2026                                    | `gpt-5` or `gpt-4.1`      |
| `gpt-4-1106-preview`                                | Mar 26, 2026, and listed again for Oct 23, 2026 | `gpt-5.6-sol`             |
| `gpt-4-turbo`, `gpt-4-turbo-2024-04-09`             | Oct 23, 2026                                    | `gpt-5.6-sol`             |
| `gpt-4`, `gpt-4-0613`                               | Oct 23, 2026                                    | `gpt-5.6-sol`             |
| `gpt-4o-2024-05-13`                                 | Oct 23, 2026                                    | `gpt-5.6-sol`             |
| Fine-tuned GPT-4 models (`ft-gpt-4`)                | Oct 23, 2026                                    | `gpt-5.6-sol` as the base |

OpenAI lists the original DevDay snapshot, `gpt-4-1106-preview`, in both the September 2025 and the April 2026 notices; either way, it has no future after October 23, 2026.

OpenAI's vocabulary is precise. A model is deprecated the moment its retirement is announced, and every deprecated model has a shutdown date after which it is no longer accessible. Legacy means it gets no more updates and will be deprecated at some point. For generally available models OpenAI promises at least six months' notice; for preview models, those with preview in the name, notice can be as short as two weeks. In some cases developers can arrange dedicated capacity through OpenAI's sales team to keep using a model after its shutdown date.

> [!WARNING]
> Moving to GPT-5 is not a long-term fix. OpenAI has already scheduled the first GPT-5 snapshot, `gpt-5-2025-08-07`, and `o3-2025-04-16` for shutdown on December 11, 2026. The Assistants API, launched alongside GPT-4 Turbo, shut down on August 26, 2026 in favour of the Responses and Conversations APIs. Whatever you migrate to, plan for the next migration.

Azure OpenAI keeps its own schedule. Under Microsoft's [lifecycle policy](https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirements), a generally available model version retires 18 months after launch, subscription owners are notified at least 60 days ahead, and a retired model answers every request with `410 Gone`. The [model retirement schedule](https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule), last updated September 21, 2026, does not list any GPT-4 or GPT-4 Turbo version. The oldest GPT-4o version it lists, 2024-05-13, retires on December 9, 2026, when Standard deployments are upgraded automatically to `gpt-5.6-sol`; provisioned deployments are not upgraded and must be migrated by hand.

## Which model should replace GPT-4 Turbo?

The substitute OpenAI's deprecations page now names is [`gpt-5.6-sol`](https://developers.openai.com/api/docs/models/gpt-5.6-sol), the top tier of the GPT-5.6 family released on July 9, 2026: a 1,050,000-token context window, up to 128,000 output tokens and a February 16, 2026 knowledge cutoff. It lists at $4 per million input tokens and $20 per million output tokens, a promotional price OpenAI says runs at least through November 21, 2026.

Since GPT-5.6, OpenAI has released GPT-6: Astra on September 3, 2026, and Sol and Luna on September 22. The [GPT-6 Sol page](https://developers.openai.com/api/docs/models/gpt-6-sol), for example, lists a 1,050,000-token context window and up to 128,000 output tokens per response, more than 30 times GPT-4 Turbo's output limit. Prices from [OpenAI's pricing page](https://developers.openai.com/api/docs/pricing) as of September 25, 2026, standard processing, for prompts up to 272,000 tokens:

| Model                    | Input, per 1M | Cached input, per 1M | Output, per 1M | Context window | Knowledge cutoff |
| ------------------------ | ------------- | -------------------- | -------------- | -------------- | ---------------- |
| `gpt-4-turbo-2024-04-09` | $10.00        | Not offered          | $30.00         | 128,000        | Dec 1, 2023      |
| `gpt-4o`                 | $2.50         | $1.25                | $10.00         | 128,000        | Oct 1, 2023      |
| `gpt-4.1`                | $2.00         | $0.50                | $8.00          | 1 million      | Jun 2024         |
| `gpt-5.6-sol`            | $4.00         | $0.40                | $20.00         | 1.05 million   | Feb 16, 2026     |
| `gpt-6-astra`            | $10.00        | $1.00                | $50.00         | 1.05 million   | Apr 30, 2026     |
| `gpt-6-sol`              | $2.00         | $0.20                | $10.00         | 1.05 million   | Apr 20, 2026     |
| `gpt-6-luna`             | $0.10         | $0.01                | $0.50          | 1.05 million   | May 18, 2026     |

Read the table with its footnotes. On GPT-6, prompts over 272,000 input tokens cost twice the input rate and 1.5 times the output rate for the whole request; Batch and Flex processing cost half the standard rate, and Fast mode, the renamed Priority processing, costs double. Regional processing adds 10% for eligible models released on or after March 5, 2026. GPT-5 lists at $1.25 input and $10 output, but its first snapshot shuts down in December.

Per-token prices do not give you the cost of a task. A reasoning model bills its hidden reasoning as output, and OpenAI's own [GPT-6 guide](https://developers.openai.com/api/docs/guides/latest-model) argues that Astra's higher per-token price is offset by using fewer output tokens per task. That is a claim to test on your own requests, not one to budget on.

If your GPT-4 Turbo workload is chat, extraction, classification or summarization, start the comparison with the cheaper tiers at low or no reasoning effort: GPT-6 Sol and Luna accept `none`, while GPT-6 Astra starts at `low`. Add Astra, or a higher effort setting, only for the cases the cheaper model fails.

## How to migrate off GPT-4 Turbo

A model change is a software change. Treat it like one, and finish well before October 23, 2026:

1. **Find every call.** Search code, configuration and environment files for the old model names. In a Git repository:

   ```bash
   git grep -nE 'gpt-4-turbo|gpt-4-0613|gpt-4-1106|gpt-4-0125|gpt-4o-2024-05-13|"gpt-4"'
   ```

   Then check what code search misses: Azure OpenAI deployment names, which you choose yourself and may not contain the model name; low-code flows; fine-tuned models built on GPT-4; and vendor products that call GPT-4 Turbo on your behalf.

2. **Build an evaluation set before you change anything.** Collect real inputs from logs and from the people who handle the work, each with what a good output must contain and a pass rule. Include typical, edge and adversarial cases, as [OpenAI's evaluation guide](https://developers.openai.com/api/docs/guides/evaluation-best-practices) recommends; the same guide lists "vibe-based evals", judging a model by whether it seems to work, as an anti-pattern. Start small and add every production failure to the set.
3. **Shortlist candidates:** OpenAI's substitute, a cheaper tier, and, if your data terms allow it, a model from another vendor.
4. **Pin the model in configuration, not in code.** Use a dated snapshot where one exists. OpenAI's [text generation guide](https://developers.openai.com/api/docs/guides/text) recommends pinning production applications to specific snapshots and building evaluation suites to watch behaviour when you upgrade. As of September 2026 the GPT-6 model pages list the alias itself, such as `gpt-6-sol`, as the default snapshot, so record the date you tested and watch the changelog.
5. **Update the integration, not only the model name.** Replace JSON mode with Structured Outputs where you parse JSON. Move tool calls from Chat Completions to the Responses API, which [GPT-6 Astra requires for tool calling](https://developers.openai.com/api/docs/changelog). Remove custom `temperature`, `top_p` and `logprobs` for Astra, which does not accept them. Raise the output limit to leave room for reasoning.
6. **Run the evaluation set on every candidate** and compare pass rate, cost per task and latency. Keep the cheapest model that passes.
7. **Roll out in stages and keep a fallback.** Send a small share of traffic first and compare it with the old model while that still runs. After the switch, rerun the evaluation set whenever the model, the prompt or the source data changes.

A minimal call with the model name taken from configuration, following the pattern in OpenAI's reasoning guide:

```javascript
import OpenAI from "openai";

const openai = new OpenAI();

const response = await openai.responses.create({
  model: process.env.OPENAI_MODEL, // the pinned model for this environment
  reasoning: { effort: "low" },
  max_output_tokens: 25000,
  input: "Classify this support email as billing, technical or other: ...",
});

console.log(response.output_text);
```

## How to compare models and vendors without trusting leaderboards

Launch posts and leaderboards answer a different question from yours: which model scored best on someone else's test, under settings the publisher chose. Two recent findings show how far that can drift from real performance.

- **Leaderboards reward selective testing.** A 2025 study of Chatbot Arena, [The Leaderboard Illusion](https://arxiv.org/abs/2504.20879), found that a handful of providers tested many private variants before release and could retract scores; it identified 27 private variants tested by Meta ahead of the Llama 4 release. It also estimated that Google and OpenAI had received 19.2% and 20.4% of all Arena data, against 29.7% for 83 open-weight models combined.
- **Benchmarks wear out.** In February 2026 [OpenAI stopped reporting SWE-bench Verified](https://openai.com/index/why-we-no-longer-evaluate-swe-bench-verified/), a coding benchmark it had created in 2024. In an audited subset of problems that models often failed, at least 59.4% had flawed tests that rejected correct solutions, and every frontier model OpenAI tested could reproduce the reference fix or problem details for some tasks, a sign they had seen them in training.

Our [Gemini explainer](https://computese.com/google-unveils-gemini-the-most-advanced-and-versatile-ai-model-yet/) shows the same problem from another angle: how the prompting method, contamination and shortcuts move a headline score. [Microsoft's Azure guidance](https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirements) makes the practical point: test candidate models on your own application and data, and weigh quality, latency and cost together instead of leaning on public benchmarks.

The comparison that predicts your results is the one you run yourself: the same evaluation set, prompts adapted to each vendor's documented style, and the same scoring rules for every candidate.

![An orange stack of test cards is copied to three model servers, and each server's answers land on a score sheet of grey check marks and crosses.](https://computese.com/images/blog/openai-unveils-gpt-4-turbo-the-most-powerful-ai-yet/eval-set.a503b69980-1536.webp)

*Run the same real cases through every candidate, and the choice rests on your work rather than someone else's benchmark.*

| What to compare | What to measure                                            | Why it matters                                                                                       |
| --------------- | ---------------------------------------------------------- | ---------------------------------------------------------------------------------------------------- |
| Quality         | Pass rate on your evaluation set, by type of case          | A leaderboard rank does not predict it                                                               |
| Cost per task   | Input, output and reasoning tokens per case, at list price | Reasoning tokens are billed as output                                                                |
| Latency         | Typical and slowest response time at your volume           | Reasoning effort and processing tier change it                                                       |
| Limits          | Context window, output limit, rate limits                  | GPT-4 Turbo's 4,096-token output limit shaped what it could do                                       |
| Data terms      | Where data is processed, retention, use for training       | On GPT-6, EU data residency is available only with standard processing                               |
| Lifecycle       | Notice period and shutdown history                         | OpenAI promises at least six months' notice for GA models; Azure retires GA versions after 18 months |

Keep the evaluation set in your own repository, not in a vendor's dashboard. OpenAI is [closing its own hosted Evals platform](https://developers.openai.com/api/docs/deprecations): existing evals become read-only on October 31, 2026, and the dashboard and API are scheduled to shut down on November 30, 2026. Test cases that live next to your code survive a change of vendor, and so do tools built on open standards such as the Model Context Protocol; our [comparison of AI frameworks](https://computese.com/latest-ai-tools-and-frameworks-a-comparative-analysis/) covers which agent SDKs support it.

If you want this done for a real workflow, our [AI automation service](https://computese.com/services/ai-automation/) builds an evaluation set from your real cases and runs it before any model, prompt or source change goes live; if the scores drop, the change does not ship. Model and prompt versions are pinned, and a fallback model covers a provider outage.

## Key terms
- **GPT-4 Turbo**: OpenAI's cheaper, long-context successor to GPT-4: a preview in November 2023, then the stable gpt-4-turbo-2024-04-09 in April 2024, with a 128,000-token context window.
- **Token**: The unit a language model reads and writes, usually a word or part of a word. API prices are quoted per million input tokens and per million output tokens.
- **Context window**: The most tokens a model can handle in one request, counting both the input you send and the output it writes.
- **Knowledge cutoff**: The date after which a model's training data has no information. A newer model does not always have a later cutoff.
- **Model snapshot**: A dated, fixed version of a model, such as gpt-4-turbo-2024-04-09. An alias, such as gpt-4-turbo, points at a snapshot and can be moved to a newer one.
- **JSON mode and Structured Outputs**: JSON mode, introduced with GPT-4 Turbo, makes the model return valid JSON. Structured Outputs, available from GPT-4o on, also makes the JSON match a schema you supply.
- **Function calling**: The model returns the name and arguments of a function your code defines, and your code runs it. Also called tool calling.
- **Reasoning model**: A model that generates hidden reasoning tokens before it answers, such as OpenAI's o-series, GPT-5 and GPT-6. Reasoning tokens use context space and are billed as output tokens.
- **Deprecation and shutdown**: OpenAI deprecates a model when it announces the retirement, and sets a shutdown date after which the model can no longer be called. A legacy model gets no updates and will be deprecated later.
- **Evaluation set**: A fixed collection of real inputs, each with what a good output must contain and a pass rule, run against every candidate model before a model or prompt change goes live.

## Common questions

### Is GPT-4 Turbo deprecated?

Yes. OpenAI's deprecations page lists gpt-4-turbo and its snapshot gpt-4-turbo-2024-04-09 for shutdown on October 23, 2026, with gpt-5.6-sol as the substitute. The older preview snapshot gpt-4-0125-preview, reached through the gpt-4-turbo-preview alias, was shut down on March 26, 2026.

### What is the difference between GPT-4 Turbo and GPT-4o?

GPT-4o, released on May 13, 2024, replaced GPT-4 Turbo as OpenAI's flagship; OpenAI said it was twice as fast, half the price and had five times the rate limits. Both have a 128,000-token context window, but GPT-4o writes up to 16,384 tokens per response against GPT-4 Turbo's 4,096, and supports Structured Outputs.

### What is the difference between GPT-4 and GPT-4 Turbo?

GPT-4 Turbo raised the context window from GPT-4's 8K or 32K tokens to 128K, moved the knowledge cutoff to April 2023 and cut the price to $0.01 per 1,000 input tokens and $0.03 per 1,000 output tokens, a third and a half of GPT-4 8K's prices. It also added JSON mode and, from April 2024, image input.

### What is GPT-4 Turbo's context window and output limit?

128,000 tokens of context, which OpenAI described at launch as more than 300 pages of text, and at most 4,096 output tokens per response. The context window counts both what you send and what the model writes.

### How much does GPT-4 Turbo cost?

As of September 2026, OpenAI lists gpt-4-turbo-2024-04-09 at $10 per million input tokens and $30 per million output tokens, or half that through the Batch API. For comparison, GPT-6 Sol lists at $2 and $10, and GPT-6 Luna at $0.10 and $0.50.

### What should replace GPT-4 Turbo?

OpenAI's deprecations page names gpt-5.6-sol as the substitute. OpenAI has since released GPT-6 Astra, Sol and Luna, and the first GPT-5 snapshot is itself due to shut down on December 11, 2026. Shortlist two or three models, including a cheaper tier and another vendor, and choose the cheapest one that passes an evaluation set built from your own cases.

## Sources
1. [New models and developer products announced at DevDay](https://openai.com/index/new-models-and-developer-products-announced-at-devday/), OpenAI
2. [Deprecations](https://developers.openai.com/api/docs/deprecations), OpenAI API documentation
3. [GPT-4 Turbo](https://developers.openai.com/api/docs/models/gpt-4-turbo), OpenAI API documentation
4. [Changelog](https://developers.openai.com/api/docs/changelog), OpenAI API documentation
5. [Hello GPT-4o](https://openai.com/index/hello-gpt-4o/), OpenAI
6. [GPT-4o](https://developers.openai.com/api/docs/models/gpt-4o), OpenAI API documentation
7. [Structured model outputs](https://developers.openai.com/api/docs/guides/structured-outputs), OpenAI API documentation
8. [GPT-4](https://developers.openai.com/api/docs/models/gpt-4), OpenAI API documentation
9. [Introducing OpenAI o1](https://openai.com/index/introducing-openai-o1-preview/), OpenAI
10. [Introducing GPT-4.1 in the API](https://openai.com/index/gpt-4-1/), OpenAI
11. [GPT-5](https://developers.openai.com/api/docs/models/gpt-5), OpenAI API documentation
12. [Reasoning models](https://developers.openai.com/api/docs/guides/reasoning), OpenAI API documentation
13. [o3 model](https://developers.openai.com/api/docs/models/o3), OpenAI API documentation
14. [Models](https://developers.openai.com/api/docs/models), OpenAI API documentation
15. [Foundry Models lifecycle and support policy](https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirements), Microsoft Learn
16. [Model retirement schedule](https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule), Microsoft Learn
17. [GPT-5.6 Sol](https://developers.openai.com/api/docs/models/gpt-5.6-sol), OpenAI API documentation
18. [GPT-6 Sol](https://developers.openai.com/api/docs/models/gpt-6-sol), OpenAI API documentation
19. [Pricing](https://developers.openai.com/api/docs/pricing), OpenAI API documentation
20. [Using GPT-6](https://developers.openai.com/api/docs/guides/latest-model), OpenAI API documentation
21. [Evaluation best practices](https://developers.openai.com/api/docs/guides/evaluation-best-practices), OpenAI API documentation
22. [Text generation](https://developers.openai.com/api/docs/guides/text), OpenAI API documentation
23. [The Leaderboard Illusion](https://arxiv.org/abs/2504.20879), arXiv (Singh et al., 2025)
24. [Why SWE-bench Verified no longer measures frontier coding capabilities](https://openai.com/index/why-we-no-longer-evaluate-swe-bench-verified/), OpenAI