# GPT-4o mini

> Source: https://aiwiki.ai/wiki/gpt_4o_mini
> Updated: 2026-11-01
> Fact-checked: 2026-11-02
> Categories: Multimodal AI, OpenAI, Small Language Models
> License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) - attribute to "AI Wiki (aiwiki.ai)"
> Cite as: AI Wiki. "GPT-4o mini." aiwiki.ai, 1 Nov 2026. https://aiwiki.ai/wiki/gpt_4o_mini
> From AI Wiki (https://aiwiki.ai), the free encyclopedia of artificial intelligence. Reuse freely with attribution.

**GPT-4o mini** is a small, low-cost multimodal [large language model](https://aiwiki.ai/wiki/large_language_model) developed by [OpenAI](https://aiwiki.ai/wiki/openai) and released on July 18, 2024, as the company's most cost-efficient model at the time. Priced at $0.15 per million input tokens and $0.60 per million output tokens, more than 60% cheaper than [GPT-3.5 Turbo](https://aiwiki.ai/wiki/gpt-3.5), it replaced GPT-3.5 Turbo as the default lightweight model in OpenAI's API and as the free-tier model in [ChatGPT](https://aiwiki.ai/wiki/chatgpt). GPT-4o mini accepts text and image inputs, has a 128,000-token [context window](https://aiwiki.ai/wiki/context_window), scores 82.0% on [MMLU](https://aiwiki.ai/wiki/mmlu), and was roughly an order of magnitude cheaper than the [frontier models](https://aiwiki.ai/wiki/frontier_models) of its era, making it the cost-efficient workhorse model of the pre-[GPT-5](https://aiwiki.ai/wiki/gpt-5) period. OpenAI described it as "our most cost-efficient small model," positioned to "expand the range of applications built with AI by making intelligence much more affordable." [1][2]

<figure class="w-56">
<img src="/images/wd-2026-10-28-openai-logo-2a509640a546-tile.png" alt="OpenAI logo" width="480" height="178" />
<figcaption>OpenAI logo. Source: <a href="https://commons.wikimedia.org/wiki/File:OpenAI_logo_2025_(wordmark).svg">Wikimedia Commons</a>, <a href="https://commons.wikimedia.org/wiki/File:OpenAI_logo_2025_(wordmark).svg#Licensing">Public domain</a>.</figcaption>
</figure>

## Overview

GPT-4o mini belongs to the GPT-4o ("omni") family but is a smaller, distinct model from the full [GPT-4o](https://aiwiki.ai/wiki/gpt_4o). OpenAI introduced it as a way to expand the range of applications that could be built affordably with AI, citing low-cost use cases such as chaining or parallelizing multiple model calls, passing large volumes of context to a model, and powering fast, real-time text interactions like customer support [chatbot](https://aiwiki.ai/wiki/chatbot)s. [1]

OpenAI did not disclose the model's parameter count, but described it as occupying roughly the same capability tier as small models such as [Gemini 1.5 Flash](https://aiwiki.ai/wiki/gemini_1_5_flash) and [Claude 3 Haiku](https://aiwiki.ai/wiki/claude_3_haiku). The company noted that the cost per token of GPT-4o mini had dropped by 99% since text-davinci-003, a less capable model released in 2022, illustrating the rapid decline in inference costs. [1][2]

At launch GPT-4o mini supported text and vision (image) inputs through the API, with OpenAI stating that support for text, image, video, and audio inputs and outputs was planned for the future. [1][4]

## What can GPT-4o mini do?

GPT-4o mini is a general-purpose model handling [natural language understanding](https://aiwiki.ai/wiki/natural_language_understanding), reasoning, math, coding, and multimodal (text-and-vision) tasks. OpenAI highlighted its strong performance in [function calling](https://aiwiki.ai/wiki/function_calling), which lets developers build applications that fetch data or take actions with external systems, as well as improved long-context performance relative to GPT-3.5 Turbo. [1]

The model uses the same tokenizer as GPT-4o, which is more efficient at handling non-English text than the tokenizer used by GPT-3.5 Turbo. [1] Its 128,000-token context window is large enough to hold roughly a book's worth of text in a single request, and it can generate up to 16,384 output tokens per request. The training data has a knowledge cutoff of October 2023. [4][5]

| Specification | Value |
| --- | --- |
| [Context window](https://aiwiki.ai/wiki/context_window) | 128,000 tokens |
| Maximum output | 16,384 tokens |
| Knowledge cutoff | October 2023 |
| Input modalities | Text, image (vision) |
| Output modalities | Text |
| Released | July 18, 2024 |

On safety, OpenAI stated that GPT-4o mini was the first model to apply its "instruction hierarchy" method, a technique designed to improve the model's ability to resist jailbreaks, [prompt injection](https://aiwiki.ai/wiki/prompt_injection)s, and system-prompt extractions. The approach helps the model prioritize developer-supplied system instructions over conflicting user instructions, reducing the effectiveness of attacks such as telling the model to "ignore all previous instructions." [17] OpenAI said this should make the model more reliable for developers deploying it at scale. [1]

## How does GPT-4o mini perform on benchmarks?

OpenAI reported that GPT-4o mini outperformed GPT-3.5 Turbo and other small models on academic benchmarks spanning reasoning, math, and coding. On textual reasoning and intelligence as measured by MMLU (Massive Multitask Language Understanding), it scored 82.0%, ahead of Gemini 1.5 Flash at 77.9% and Claude 3 Haiku at 73.8%. On the MGSM benchmark for mathematical reasoning, it scored 87.0%, compared with 75.5% for Gemini 1.5 Flash and 71.7% for Claude 3 Haiku. On [HumanEval](https://aiwiki.ai/wiki/humaneval), which measures coding, it scored 87.2%, ahead of Gemini 1.5 Flash at 71.5% and Claude 3 Haiku at 75.9%. [1][6]

In multimodal reasoning measured by MMMU (Massive Multi-discipline Multimodal Understanding), GPT-4o mini scored 59.4%, compared with 56.1% for Gemini 1.5 Flash and 50.2% for Claude 3 Haiku. OpenAI also reported scores of 70.2% on the MATH benchmark and 79.7% on DROP, a reading-comprehension and reasoning benchmark. [7][15] For comparison, OpenAI's simple-evals table lists Gemini 1.5 Flash at 40.9 on MATH and 78.4 on DROP, and Anthropic reported Claude 3 Haiku at 38.9% on MATH and 78.4 on DROP. [15][16]

| Benchmark | GPT-4o mini | Gemini 1.5 Flash | Claude 3 Haiku |
| --- | --- | --- | --- |
| MMLU (reasoning) | 82.0% | 77.9% | 73.8% |
| MGSM (math) | 87.0% | 75.5% | 71.7% |
| [HumanEval](https://aiwiki.ai/wiki/humaneval) (coding) | 87.2% | 71.5% | 75.9% |
| MMMU (multimodal) | 59.4% | 56.1% | 50.2% |
| MATH | 70.2% | 40.9% | 38.9% |
| DROP | 79.7% | 78.4 (F1) | 78.4 (F1) |

OpenAI further noted that, at the time of launch, an earlier version of GPT-4o mini ranked above [GPT-4](https://aiwiki.ai/wiki/gpt-4) Turbo (the 01-25 snapshot) on user chat preferences in the LMSYS [Chatbot Arena](https://aiwiki.ai/wiki/lmsys_chatbot_arena) leaderboard. [1][6]

## How much does GPT-4o mini cost?

GPT-4o mini launched at $0.15 per million input tokens and $0.60 per million output tokens (15 cents and 60 cents, respectively). OpenAI described this as more than 60% cheaper than GPT-3.5 Turbo, whose pricing had been $0.50 per million input tokens and $1.50 per million output tokens. [1][18]

| Model | Input (per 1M tokens) | Output (per 1M tokens) |
| --- | --- | --- |
| GPT-4o mini | $0.15 | $0.60 |
| GPT-3.5 Turbo | $0.50 | $1.50 |

The dramatically lower price was central to OpenAI's positioning of the model, enabling cost-sensitive workloads that would have been impractical with larger or older models. [1][3]

## When was GPT-4o mini released and where was it available?

GPT-4o mini became available on July 18, 2024 through the [OpenAI API](https://aiwiki.ai/wiki/openai_api) (Assistants API, Chat Completions API, and Batch API) and immediately replaced GPT-3.5 Turbo in ChatGPT for Free, Plus, and Team users. As OpenAI put it, "Free, Plus and Team users will be able to access GPT-4o mini starting today, in place of GPT-3.5." Enterprise users gained access the following week. [1][2]

On July 23, 2024, OpenAI made fine-tuning available for GPT-4o mini, initially to developers on usage tiers 4 and 5, saying it planned to expand access to all tiers. That expansion happened within a month: when OpenAI opened fine-tuning for the full-size GPT-4o on August 20, 2024, it stated that GPT-4o mini fine-tuning was available to all developers on all paid usage tiers. [14] As a limited-time offer, the company provided the first 2 million training tokens per day for free through September 23, 2024. On May 8, 2026, OpenAI said it was winding down its fine-tuning platform, closing it to new users while keeping existing fine-tuned models available until their base models are deprecated. [14] This announcement came shortly after Meta released its open-source [Llama 3.1](https://aiwiki.ai/wiki/llama_3_1) models, amid intense competition in the small-model market. [8][9]

GPT-4o mini was also offered through [Microsoft](https://aiwiki.ai/wiki/microsoft)'s [Azure OpenAI Service](https://aiwiki.ai/wiki/azure_openai). Although GPT-4o mini replaced GPT-3.5 Turbo as OpenAI's recommended small model, GPT-3.5 Turbo remained available through the API for existing applications. [4]

## What replaced GPT-4o mini?

GPT-4o mini served as OpenAI's default budget model for roughly ten months before being superseded. On April 14, 2025, OpenAI released the [GPT-4.1](https://aiwiki.ai/wiki/gpt-4.1) family, including [GPT-4.1 mini](https://aiwiki.ai/wiki/gpt_4_1_mini) and [GPT-4.1 nano](https://aiwiki.ai/wiki/gpt_4_1_nano); GPT-4.1 mini replaced GPT-4o mini as the fallback model for all ChatGPT users beginning May 14, 2025. For reasoning-heavy workloads, OpenAI's o-series small model o4-mini (released April 2025) provided a [chain-of-thought](https://aiwiki.ai/wiki/chain_of_thought) alternative in the same low-cost tier. GPT-4o mini's lineage continued with GPT-5 mini, released August 7, 2025 alongside GPT-5, which became the resource-efficient fallback for free ChatGPT users. [12][19][20]

| Model | Released | Role |
| --- | --- | --- |
| GPT-3.5 Turbo | 2023 | Prior default budget model |
| GPT-4o mini | July 18, 2024 | Cost-efficient multimodal workhorse |
| [GPT-4.1 mini](https://aiwiki.ai/wiki/gpt_4_1_mini) | April 14, 2025 | Replaced GPT-4o mini in [ChatGPT](https://aiwiki.ai/wiki/chatgpt) (May 14, 2025) |
| o4-mini | April 2025 | Small reasoning (o-series) model |
| [GPT-5](https://aiwiki.ai/wiki/gpt-5) mini | August 7, 2025 | GPT-5-era budget fallback |

On February 13, 2026, OpenAI retired GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini from the ChatGPT consumer interface, consolidating around the GPT-5 family, though API deprecation schedules for individual models run separately. [13] On October 14, 2026, OpenAI notified developers that GPT-4o mini (gpt-4o-mini-2024-07-18), along with GPT-4o, GPT-4.1, and GPT-4.1 mini, would be removed from the API on January 14, 2027, naming GPT-6 Luna as the recommended replacement. [21]

## Reception

GPT-4o mini was widely covered as a significant move to lower the cost of capable AI. TechCrunch described it as a "smaller and cheaper" model that powered ChatGPT and replaced GPT-3.5 Turbo as OpenAI's smallest offering, quoting Artificial Analysis's measurement of a median output speed of 202 [tokens per second](https://aiwiki.ai/wiki/tokens_per_second), more than twice as fast as GPT-4o and GPT-3.5 Turbo. [2] VentureBeat characterized it as a "smaller, much cheaper multimodal AI model" and emphasized its competitive benchmark results against rival small models. [4]

Commentators framed the launch as part of an escalating price and capability race among providers of small, fast models, including [Google](https://aiwiki.ai/wiki/google)'s Gemini 1.5 Flash and [Anthropic](https://aiwiki.ai/wiki/anthropic)'s Claude 3 Haiku, with OpenAI's free fine-tuning offer interpreted as a direct response to Meta's Llama 3.1 release. [8][9] Outlets such as TechRadar covered the launch under the framing of bidding "goodbye" to GPT-3.5, reflecting the model's role in superseding the older default. [11]

## What are GPT-4o mini's limitations?

GPT-4o mini is a small model and is less capable than larger frontier models, including the full GPT-4o, on the most demanding reasoning, knowledge, and multimodal tasks; OpenAI positioned it for cost-efficient and high-volume workloads rather than maximum capability. Its knowledge is fixed at an October 2023 cutoff, so it lacks awareness of events after that date unless supplemented with external context. [4][5]

At launch the model's multimodal support was limited to text and image inputs and text outputs; the audio and video capabilities that OpenAI said were planned were not available at release. [1][4] As with other large language models, GPT-4o mini can produce inaccurate or fabricated information ("hallucinations"), and OpenAI's instruction-hierarchy safety method reduces but does not entirely eliminate the risk of jailbreaks and prompt-injection attacks. [1]

## References

1. OpenAI. "GPT-4o mini: advancing cost-efficient intelligence." July 18, 2024. https://openai.com/index/gpt-4o-mini-advancing-cost-efficient-intelligence/
2. Zeff, Maxwell. "OpenAI unveils GPT-4o mini, a smaller and cheaper AI model." TechCrunch, July 18, 2024. https://techcrunch.com/2024/07/18/openai-unveils-gpt-4o-mini-a-small-ai-model-powering-chatgpt/
3. Campus Technology. "OpenAI Launches Slimmer, Cheaper GPT-4o Mini." July 18, 2024. https://campustechnology.com/articles/2024/07/18/openai-launches-slimmer-cheaper-gpt-4o-mini.aspx
4. Franzen, Carl. "OpenAI unveils GPT-4o mini, a smaller and much cheaper multimodal AI model." VentureBeat, July 18, 2024. https://venturebeat.com/ai/openai-unveils-gpt-4o-mini-a-smaller-much-cheaper-multimodal-ai-model
5. DocsBot AI. "OpenAI's GPT-4o Mini - AI Model Details." https://docsbot.ai/models/gpt-4o-mini
6. DataCamp. "What Is GPT-4o Mini? How It Works, Use Cases, API & More." https://www.datacamp.com/blog/gpt-4o-mini
7. LLM Stats. "GPT-4o mini Benchmarks, Pricing & Context Window." https://llm-stats.com/models/gpt-4o-mini-2024-07-18
8. OpenAI Developers (@OpenAIDevs). "Customize GPT-4o mini for your application with fine-tuning." X, July 23, 2024. https://x.com/OpenAIDevs/status/1815836887631946015
9. Nunez, Michael. "AI arms race escalates: OpenAI offers free GPT-4o Mini fine-tuning to counter Meta's Llama 3.1 release." VentureBeat, July 23, 2024. https://venturebeat.com/ai/ai-arms-race-escalates-openai-offers-free-gpt-4o-mini-fine-tuning-to-counter-metas-llama-3-1-release
10. Microsoft Azure. "OpenAI's fastest model, GPT-4o mini is now available on Azure AI." https://azure.microsoft.com/en-us/blog/openais-fastest-model-gpt-4o-mini-is-now-available-on-azure-ai/
11. TechRadar. "Goodbye GPT-3.5, OpenAI's new GPT-4o mini AI model is all about compact power." July 19, 2024. https://www.techradar.com/computing/artificial-intelligence/goodbye-gpt-35-openais-new-gpt-4o-mini-ai-model-is-all-about-compact-power
12. OpenAI. "Introducing GPT-4.1 in the API." April 14, 2025. https://openai.com/index/gpt-4-1/
13. OpenAI. "Retiring GPT-4o, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini in ChatGPT." 2026. https://openai.com/index/retiring-gpt-4o-and-older-models/
14. OpenAI. "Fine-tuning now available for GPT-4o." August 20, 2024. https://openai.com/index/gpt-4o-fine-tuning/
15. OpenAI. "simple-evals" (benchmark results table). GitHub. https://github.com/openai/simple-evals
16. Anthropic. "Introducing the next generation of Claude." March 4, 2024. https://www.anthropic.com/news/claude-3-family
17. Robison, Kylie. "OpenAI's latest model will block the 'ignore all previous instructions' loophole." The Verge, July 19, 2024. https://www.theverge.com/2024/7/19/24201414/openai-chatgpt-gpt-4o-prompt-injection-instruction-hierarchy
18. OpenAI. "New embedding models and API updates." January 25, 2024. https://openai.com/index/new-embedding-models-and-api-updates/
19. Disotto, John-Anthony. "OpenAI just gave ChatGPT users a huge free upgrade, 4.1 mini is available today." TechRadar, May 15, 2025. https://www.techradar.com/computing/artificial-intelligence/openai-just-gave-chatgpt-users-a-huge-free-upgrade-4-1-mini-is-available-today
20. OpenAI. "Introducing GPT-5." August 7, 2025. https://openai.com/index/introducing-gpt-5/
21. OpenAI. "Deprecations." OpenAI API documentation. https://developers.openai.com/api/docs/deprecations

