# GPT-4 Turbo ![Organization](https://img.shields.io/badge/Org-OpenAI-412991) ![Context](https://img.shields.io/badge/Context-128K%20tokens-1f6feb) ![Pricing](https://img.shields.io/badge/Price-%2410%20%2F%20%2430%20per%201M-2da44e) ![License](https://img.shields.io/badge/License-Proprietary-d29922) ![Modality](https://img.shields.io/badge/Modality-Text%20%2B%20Vision-8957e5) ![Released](https://img.shields.io/badge/Released-Apr%202024-555) > **Quick answer:** GPT-4 Turbo (API: gpt-4-turbo-2024-04-09) is OpenAI's April 2024 flagship — the last major GPT-4 release before GPT-4o. It scores 97.0% on GSM8K, 86.4% on MMLU, 9.32/10 on MT-Bench, and 28.7% on LiveCodeBench. With a 128K context window at $10/$30 per 1M tokens, it was state-of-the-art at launch but has since been superseded by GPT-4o and GPT-5.x. ## At a Glance **Where GPT-4 Turbo leads** - 97.0% GSM8K — top-tier grade-school math - 86.4% MMLU — strong academic knowledge - 9.32/10 MT-Bench — excellent instruction following - 50.0% AlpacaEval 2 — strong open-ended generation quality - 72.6% MATH — competitive competition mathematics - Stable, well-tested API for existing integrations **Where it lags** - Superseded by GPT-4o, GPT-4.1, and GPT-5.x across all benchmarks - 28.7% LiveCodeBench — well below current frontier models - $10/$30 per 1M tokens — expensive relative to far superior current alternatives - 128K context window vs 1M for GPT-4.1 and Gemini 2.5 Pro - Knowledge cutoff December 2023 **Best for:** Legacy production integrations validated on gpt-4-turbo-2024-04-09; tasks where the specific model version is contractually or compliance-required. ## What GPT-4 Turbo Is GPT-4 Turbo (gpt-4-turbo-2024-04-09) was OpenAI's highest-capability model from November 2023 through May 2024, when GPT-4o was released. It introduced a 128K context window (4× the 32K limit of GPT-4), vision support, and improved instruction following over the original GPT-4. The April 2024 version (gpt-4-turbo-2024-04-09) was notable for achieving near-human performance on MT-Bench (9.32/10) and was widely used as a baseline for LLM evaluation. Its MMLU score of 86.4% and GSM8K score of 97.0% were considered frontier results at launch. For teams still using GPT-4 Turbo, migration to GPT-4o or GPT-4.1 is strongly recommended: both offer better performance at substantially lower cost. GPT-4.1 specifically provides a 1M context window, better benchmark scores, and costs $2/$8 vs GPT-4 Turbo's $10/$30 per 1M tokens. ## Specifications | Field | Value | |---|---| | Organization | OpenAI | | Parameters | Undisclosed | | Context window | 128,000 tokens | | Max output | 4,096 tokens | | API model ID | gpt-4-turbo-2024-04-09 | | License | Proprietary (API only) | | Release date | April 9, 2024 | | Knowledge cutoff | December 2023 | | Modality | Text + Vision (multimodal) | ## Pricing | | Input (per 1M tokens) | Output (per 1M tokens) | |---|---|---| | OpenAI API | $10.00 | $30.00 | Pricing per [OpenAI pricing page](https://openai.com/api/pricing/). Note: GPT-4.1 provides better performance at $2/$8 per 1M. ## Context Window GPT-4 Turbo has a **128,000-token context window** — roughly **90 pages** of text in a single request. This is sufficient for most document tasks but is significantly smaller than GPT-4.1 (1M tokens) or Gemini 2.5 Pro (1M tokens). ## Public Benchmark Scores | Benchmark | Score | Source | Date | |---|---|---|---| | [GSM8K](/benchmarks/gsm8k) | 97.0% | Benchgen evaluation | 2024 | | [MMLU](/benchmarks/mmlu) | 86.4% | Benchgen evaluation | 2024 | | [MATH](/benchmarks/math) | 72.6% | Benchgen evaluation | 2024 | | [MT-Bench](/benchmarks/mt-bench) | 9.32 / 10 | Benchgen evaluation | 2024 | | [AlpacaEval 2](/benchmarks/alpacaeval-2) | 50.0% | Benchgen evaluation | 2024 | | [LiveCodeBench](/benchmarks/livecodebench) | 28.7% | Benchgen evaluation | 2024 | ## GPT-4 Turbo vs Alternatives | Model | Context | MMLU | LiveCodeBench | Price (in/out per 1M) | |---|---|---|---|---| | **GPT-4 Turbo** | 128K | 86.4% | 28.7% | $10 / $30 | | [GPT-4.1](/models/openai/gpt-4-1) | 1M | — | — | $2 / $8 | | [GPT-4o](/models/openai/gpt-4o) | 128K | — | — | $2.50 / $10 | | [o4-mini](/models/openai/o4-mini) | 200K | — | 74.2% | $1.10 / $4.40 | GPT-4 Turbo is uniformly worse value than its successors: GPT-4.1 offers 5× more context and better scores at 5× lower cost. For any new deployment, GPT-4.1 or GPT-4o is the recommended replacement. ## Use GPT-4 Turbo via API ```python from openai import OpenAI client = OpenAI(api_key="YOUR_API_KEY") response = client.chat.completions.create( model="gpt-4-turbo-2024-04-09", messages=[{"role": "user", "content": "Summarise this research paper..."}], ) print(response.choices[0].message.content) ``` ## Related Models & Benchmarks - Compare: [GPT-4o](/models/openai/gpt-4o) · [GPT-4.1](/models/openai/gpt-4-1) · [o4-mini](/models/openai/o4-mini) - Benchmarks: [GSM8K](/benchmarks/gsm8k) · [MMLU](/benchmarks/mmlu) · [MT-Bench](/benchmarks/mt-bench) · [LiveCodeBench](/benchmarks/livecodebench) ## Frequently Asked Questions

What is GPT-4 Turbo?

GPT-4 Turbo (gpt-4-turbo-2024-04-09) is OpenAI's April 2024 GPT-4 flagship, featuring a 128K context window and vision support. It has been superseded by GPT-4o and GPT-4.1 for new deployments.

What is GPT-4 Turbo's context window?

GPT-4 Turbo supports a 128,000-token context window with up to 4,096 output tokens.

How much does GPT-4 Turbo cost?

GPT-4 Turbo costs $10.00 per 1M input tokens and $30.00 per 1M output tokens. GPT-4.1 provides better performance at $2/$8 per 1M and is the recommended migration target.

Should I migrate from GPT-4 Turbo?

Yes. GPT-4.1 provides better benchmark scores, a 1M context window (vs 128K), and costs 80% less than GPT-4 Turbo. For most use cases, GPT-4o or GPT-4.1 are the correct migration targets.

What is GPT-4 Turbo's knowledge cutoff?

GPT-4 Turbo's training knowledge cutoff is December 2023.

--- *Specs and scores sourced from OpenAI's official GPT-4 Turbo announcement (April 2024) and Benchgen evaluations. Pricing cited to the OpenAI pricing page. Last updated 2026-07-24.*

GPT-4 Turbo — Files