ChatGPT: Updated list of available models, capabilities, and pricing (August 2025)

OpenAI’s catalog of available ChatGPT models has expanded significantly, covering GPT-5, GPT-4 variants, the o-series reasoning models, o4-mini multimodal options, research-preview releases like GPT-4.5, and lightweight on-device deployments optimized for mobile and edge inference. The offering now balances reasoning depth, multimodality, latency, and cost, giving developers and enterprise teams flexibility when integrating ChatGPT into workflows.

GPT-5 models redefine flagship capabilities.

The GPT-5 family marks OpenAI’s most advanced release, combining multimodal input (text, image, and audio), improved reasoning, and agentic orchestration for complex tasks. It introduces enhanced context handling up to 256 000 tokens, extending usability for research, analytics, and enterprise-scale datasets.

Key enhancements introduced with GPT-5.

• Expanded multimodal support with better image and table parsing.

• Planning and decision-making improvements for multi-step workflows.

• Improved consistency in responses across large datasets.

• Full availability across ChatGPT Plus, Team, and Enterprise subscriptions.

• Wider tool integration support for structured outputs and function calls.

Model	Context Window	Use Case	Pricing Status
GPT-5	256 000 tokens	Flagship multimodal reasoning	Pending
GPT-5 Pro	256 000 tokens	Deeper structured orchestration	Enterprise-tier uplift

GPT-4o models continue leading in speed and multimodality.

The GPT-4o generation, including its smaller variants, remains the primary production family for fast, cost-effective multimodal experiences. These models enable text, image, and table-based tasks while maintaining high performance for real-time chat.

Model	Context Window	Latency	Main Use	Pricing (1 000 tokens)
GPT-4o	128 000 tokens	Fast	Balanced multimodal workflows	$0.18 / $0.54
GPT-4o Mini	64 000 tokens	Very low	Lightweight chat, summarization	$0.09 / $0.27
GPT-4o Nano 20B	8 000 tokens	Ultra-low	On-device inference kit	Licensed OEM
GPT-4o Vision-Assist	64 000 tokens	Mid	Image-to-JSON structured parsing	1.2× GPT-4o

Details on Nano and Vision-Assist capabilities.

The GPT-4o Nano 20B model targets edge deployments with full local inference on supported hardware such as Pixel 10 devices, while the Vision-Assist beta supports table extraction, structured image parsing, and field-level labeling, ideal for document-heavy workflows.

GPT-4.1 and GPT-4 Turbo balance cost and performance.

GPT-4.1 bridges high-precision reasoning and affordable usage for large-scale contexts up to 1 000 000 tokens, useful for developers needing fine-grained control without sacrificing reliability.

GPT-4 Turbo, meanwhile, remains attractive for workloads needing stable performance at a lower cost, having received a recent price drop.

Model	Context Window	Primary Focus	Pricing (1 000 tokens)
GPT-4.1	1 000 000 tokens	Coding, analysis, research	$0.24 / $0.48
GPT-4.1 Mini	512 000 tokens	Middle-tier workflows	$0.12 / $0.32
GPT-4 Turbo	128 000 tokens	Backward compatibility	$0.24 / $0.48

The o-series models redefine structured reasoning.

OpenAI’s o-series introduces a new generation of models optimized for multi-step reasoning, chain-of-thought orchestration, and structured decision-making. Unlike standard GPT models, these variants prioritize analytical depth over speed, allowing developers and enterprises to handle more complex workflows with controlled token budgeting.

Key improvements introduced with o-series models.

• Adaptive reasoning levels configurable per request, supporting minimal, low, medium, and high effort modes.

• Long-context reasoning up to 200,000 tokens with better consistency across extended analytical chains.

• Native support for function calling and tool-driven orchestration.

• Optimized pricing tiers to balance cost, latency, and reasoning depth.

Model	Context Window	Reasoning Effort	Use Case	Pricing (1,000 tokens)
o1	200,000 tokens	Configurable	Advanced research, coding, data modeling	$2.00 / $8.00
o1-mini	128,000 tokens	Medium	Cost-efficient structured tasks	$1.10 / $4.40
o1-pro	200,000 tokens	High	Deep technical orchestration	$20.00 / $80.00
o3	200,000 tokens	Configurable	Balanced reasoning + multimodal support	$2.00 / $8.00
o3-mini	200,000 tokens	Low-medium	High-volume production workloads	$1.10 / $4.40
o3-mini-high	200,000 tokens	High	Extended chain-of-thought tasks	$1.10 / $4.40
o3-pro	200,000 tokens	Maximum	Competitive research, multi-agent tasks	$20.00 / $80.00

These models are especially relevant for enterprises deploying agentic multi-step reasoning pipelines or applications requiring high output quality with predictable orchestration costs.

The o4-mini series focuses on lightweight multimodal reasoning.

The o4-mini models expand the OpenAI catalog with compact, multimodal-optimized architectures that balance speed, reasoning power, and cost. They offer full support for images, text, and table parsing while maintaining sub-second latency for smaller workloads.

Key enhancements introduced with o4-mini models.

• Improved vision-to-JSON structured parsing.

• Lowered inference costs while retaining access to high reasoning effort modes.

• Support for both tool invocation and real-time multimodal workflows.

Model	Context Window	Reasoning Effort	Capabilities	Pricing (1,000 tokens)
o4-mini	128,000 tokens	Configurable	Balanced multimodal processing	$1.10 / $4.40
o4-mini-high	128,000 tokens	High	Deeper reasoning in edge tasks	$1.10 / $4.40

This makes the o4-mini line especially suitable for enterprise dashboards, structured analytics, and real-time document parsing in environments where latency constraints are critical.

GPT-4.5 expands creativity and pattern recognition.

GPT-4.5 launched as a research preview, designed for users who need improved pattern detection, creative synthesis, and faster context adaptation across long sessions. Unlike GPT-4 Turbo, it focuses less on transactional chat performance and more on exploratory ideation.

Model	Context Window	Primary Use	Pricing (1,000 tokens)	Availability
GPT-4.5	128,000 tokens	Brainstorming, complex ideation, high-context summarization	$75.00 / $150.00	Research preview

Currently, GPT-4.5 is available under a limited access tier via ChatGPT Pro and Azure Foundry, making it more suitable for R&D teams rather than mainstream enterprise adoption.

GPT-5 mini and GPT-5 nano enable scalable multimodal inference.

In addition to the flagship GPT-5 and GPT-5 Pro, OpenAI has introduced lighter GPT-5 derivatives optimized for speed and affordability while retaining multimodal capabilities.

Model	Context Window	Latency	Use Case	Pricing (1,000 tokens)
GPT-5	256,000 tokens	Mid-range	Enterprise-scale reasoning	$1.25 / $10.00
GPT-5 mini	128,000 tokens	Low	High-volume lightweight workflows	$0.25 / $2.00
GPT-5 nano	64,000 tokens	Ultra-low	Mobile inference, edge integrations	$0.05 / $0.40

These smaller GPT-5 variants automatically integrate into OpenAI’s routing system, allowing developers to scale workloads dynamically without manually switching between model endpoints.

GPT-3.5 Turbo transitions to legacy but remains relevant.

While GPT-3.5 Turbo is scheduled for deprecation in February 2026, it still supports fine-tuning projects and lightweight chat experiences where cost efficiency is the primary driver.

Model	Context Window	Availability	Pricing (1 000 tokens)
GPT-3.5 Turbo	16 000 tokens	Legacy until Feb 2026	$0.005 / $0.015

Open-weight GPT OSS models enable local deployment.

OpenAI has also introduced GPT OSS models with 20B and 120B parameters, providing developers with self-hosted capabilities for scenarios requiring full control of inference, privacy, and latency.

Model	Context Window	Parameters	Deployment Type	License
gpt-oss-20B	8 000 tokens	20B	Local inference	MIT-style
gpt-oss-120B	8 000 tokens	120B	Private hosting	MIT-style

____________

FOLLOW US FOR MORE.

DATA STUDIOS

datastudios.org