GPT-5.5 is OpenAI’s frontier model designed for complex professional workloads, building on GPT-5.4 with stronger reasoning, higher reliability, and improved token efficiency on hard tasks. It features a 1M+ token context window (922K input, 128K output) with support for text and image inputs, enabling large-scale reasoning, coding, and multimodal workflows within a single system.
gpt-image-2 (officially launched as part of ChatGPT Images 2.0) is OpenAI's most advanced image synthesis engine, integrated directly into the GPT-5.4 backbone. Unlike previous pipeline-based models, it features Agentic Reasoning, allowing the model to "think," plan, and search the web for real-time context before rendering a single pixel. It delivers near-perfect text rendering (~99% accuracy) across multiple languages including CJK scripts. With its 4K resolution support and advanced spatial intelligence, gpt-image-2 can generate up to 8 coherent images in a single request, maintaining perfect character and object continuity. In 2026, it is the primary choice for professional UI/UX prototyping, cinematic assets, and high-fidelity branding.
gpt-image-2 (officially launched as part of ChatGPT Images 2.0) is OpenAI's most advanced image synthesis engine, integrated directly into the GPT-5.4 backbone. Unlike previous pipeline-based models, it features Agentic Reasoning, allowing the model to "think," plan, and search the web for real-time context before rendering a single pixel. It delivers near-perfect text rendering (~99% accuracy) across multiple languages including CJK scripts. With its 4K resolution support and advanced spatial intelligence, gpt-image-2 can generate up to 8 coherent images in a single request, maintaining perfect character and object continuity. In 2026, it is the primary choice for professional UI/UX prototyping, cinematic assets, and high-fidelity branding.
GPT-5.4 is OpenAI’s latest frontier model, unifying the Codex and GPT lines into a single system. It features a 1M+ token context window (922K input, 128K output) with support for text and image inputs, enabling high-context reasoning, coding, and multimodal analysis within the same workflow. The model delivers improved performance in coding, document understanding, tool use, and instruction following. It is designed as a strong default for both general-purpose tasks and software engineering, capable of generating production-quality code, synthesizing information across multiple sources, and executing complex multi-step workflows with fewer iterations and greater token efficiency.
OpenAI's latest flagship image model. Moving beyond traditional diffusion artifacts, it uses a full Transformer architecture for "pixel-level reasoning." The core evolution lies in Identity Locking and physical lighting consistency. It supports extreme instruction adherence (e.g., placing objects at specific coordinates) and precise editing, such as changing clothing or lighting without altering the face or background.
OpenAI's latest flagship image model. Moving beyond traditional diffusion artifacts, it uses a full Transformer architecture for "pixel-level reasoning." The core evolution lies in Identity Locking and physical lighting consistency. It supports extreme instruction adherence (e.g., placing objects at specific coordinates) and precise editing, such as changing clothing or lighting without altering the face or background.
OpenAI's premium video-audio model featuring synchronized 1080p output, simulation-grade physics, and native 25-second coherence. It supports licensed "Disney Cameos" and professional-grade temporal stability, ideal for cinematic, marketing, and commercial production. | Parameter | Type | Required | Group | Allowed Values | |-----------|--------------|-----------|------------|--------------------| | seconds | enum<string> | Optional | Default | 10, 15, 25 | | | | | Reverse | 10, 15, 25 | | | | | Official | 4, 8, 12 | | | | | Enterprise | 4, 8, 12 |
OpenAI's flagship video-and-audio generation model. It features synchronized dialogue, 1080p HD quality, and improved physics realism. With the "Cameo" feature and a $1B Disney partnership, users can create cinematic videos up to 25 seconds long with licensed characters. | Parameter | Type | Required | Group | Allowed Values | |-----------|--------------|-----------|------------|--------------------| | seconds | enum<string> | Optional | Default | 10, 15 | | | | | Reverse | 10, 15 | | | | | Official | 4, 8, 12 | | | | | Enterprise | 4, 8, 12 |
gpt-oss-120bis our most powerful open-weight model, which fits into a single H100 GPU (117B parameters with 5.1B active parameters). Download gpt-oss-120b on HuggingFace. Key features Permissive Apache 2.0 license: Build freely without copyleft restrictions or patent risk—ideal for experimentation, customization, and commercial deployment. Configurable reasoning effort: Easily adjust the reasoning effort (low, medium, high) based on your specific use case and latency needs. Full chain-of-thought: Gain complete access to the model's reasoning process, facilitating easier debugging and increased trust in outputs. Fine-tunable: Fully customize models to your specific use case through parameter fine-tuning. Agentic capabilities: Use the models' native capabilities for function calling, web browsing, Python code execution, and structured outputs.
gpt-oss-20b is our medium-sized open-weight model for low latency, local, or specialized use-cases (21B parameters with 3.6B active parameters). Download gpt-oss-20b on HuggingFace. Key features Permissive Apache 2.0 license: Build freely without copyleft restrictions or patent risk—ideal for experimentation, customization, and commercial deployment. Configurable reasoning effort: Easily adjust the reasoning effort (low, medium, high) based on your specific use case and latency needs. Full chain-of-thought: Gain complete access to the model's reasoning process, facilitating easier debugging and increased trust in outputs. Fine-tunable: Fully customize models to your specific use case through parameter fine-tuning. Agentic capabilities: Use the models' native capabilities for function calling, web browsing, Python code execution, and structured outputs.
A cost-optimized image generation and editing model released by OpenAI in July 2025. As the entry-level member of the GPT Image family, it shares the same native multimodal Transformer architecture as the flagship models. While sacrificing some top-end fidelity, it excels in high-throughput production, e-commerce batch editing, and rapid prototyping with significantly lower latency.
A landmark image model released by OpenAI in April 2025. It is the first to unify visual understanding and generation within a single Transformer architecture. By natively supporting mixed text and image inputs, it excels in image generation, high-precision inpainting, identity-preserved editing, and complex typography.
gpt-4.1 is the flagship variant of OpenAI's GPT-4.1 family, officially designated as the "Smartest non-reasoning model." It marks a significant advancement over GPT-4o, with substantial improvements in coding reliability and instruction-following precision. Its standout technical breakthrough is the 1-million-token context window, enabling the processing of entire monorepos, extensive legal documents, or massive event logs in a single pass. The model is trained for high "literal adherence," following complex formatting and logic far more accurately than its predecessors. It also introduces optimized code-diffing capabilities, allowing for precise edits that reduce both cost and latency for developers.
gpt-4.1-nano is the most lightweight and cost-efficient variant in OpenAI's GPT-4.1 family. As the first model from OpenAI to carry the "Nano" designation, it is engineered for near-instant latency and high-volume throughput. Its standout feature is the 1-million-token context window, allowing users to process massive datasets or entire code repositories in a single prompt. Despite its small scale, it outperforms GPT-4o mini on core benchmarks and shows significant improvements in literal instruction following. It is ideal for high-speed tasks such as classification, autocomplete, and acting as a pre-processor for long-context retrieval where deep reasoning steps are not required.
gpt-4o-2024-11-20 is an optimized snapshot of the GPT-4o series. Built on the "Omni" architecture, it natively integrates text, audio, image, and video processing. Compared to earlier snapshots like 0513 or 0806, this version delivers a noticeable boost in creative writing, producing more natural, human-like prose with fewer repetitive patterns. It also shows improved robustness in coding and superior instruction-following, particularly for generating complex structured data (JSON). It remains the industry's most versatile, high-speed, and cost-efficient flagship model for general-purpose AI applications.
The 2024-11-20 version of GPT-4o offers a leveled-up creative writing ability with more natural, engaging, and tailored writing to improve relevance & readability. It’s also better at working with uploaded files, providing deeper insights & more thorough responses. GPT-4o ("o" for "omni") is OpenAI's latest AI model, supporting both text and image inputs with text outputs. It maintains the intelligence level of GPT-4 Turbo while being twice as fast and 50% more cost-effective. GPT-4o also offers improved performance in processing non-English languages and enhanced visual capabilities.
text-embedding-3-large is OpenAI's most capable embedding model for both english and non-english tasks. Embeddings are a numerical representation of text that can be used to measure the relatedness between two pieces of text. Embeddings are useful for search, clustering, recommendations, anomaly detection, and classification tasks.
text-embedding-3-small is OpenAI's improved, more performant version of the ada embedding model. Embeddings are a numerical representation of text that can be used to measure the relatedness between two pieces of text. Embeddings are useful for search, clustering, recommendations, anomaly detection, and classification tasks.
text-embedding-3-large is OpenAI's most capable embedding model for both english and non-english tasks. Embeddings are a numerical representation of text that can be used to measure the relatedness between two pieces of text. Embeddings are useful for search, clustering, recommendations, anomaly detection, and classification tasks.