OpenAI releases ChatGPT Images 2.0 with 3840x2160 resolution at $30 per 1M output tokens
OpenAI released ChatGPT Images 2.0, pricing output tokens at $30 per million with maximum resolution of 3840x2160 pixels. CEO Sam Altman claims the improvement from gpt-image-1 to gpt-image-2 equals the jump from GPT-3 to GPT-5.
OpenAI releases ChatGPT Images 2.0 with 3840x2160 resolution at $30 per 1M output tokens
OpenAI released ChatGPT Images 2.0, its latest image generation model. CEO Sam Altman claims the leap from gpt-image-1 to gpt-image-2 is equivalent to jumping from GPT-3 to GPT-5 in text models.
Pricing and specifications
The model charges $30 per million output tokens. A maximum resolution 3840x2160 image generated 13,342 output tokens, costing approximately $0.40. The model includes an outputQuality parameter that can be set to "high" for improved results.
The OpenAI Python client library has not yet been updated to include gpt-image-2 in its validated model list, but the model ID can be used directly with the API.
Comparative testing results
Developer Simon Willison tested the model using a "Where's Waldo" style prompt: creating a busy scene where viewers must find a raccoon holding a ham radio.
gpt-image-1 (baseline): Failed to include an identifiable raccoon in the generated scene. Claude Opus 4.7 could not locate a raccoon when analyzing the output.
Google Nano Banana 2: Successfully generated a clearly visible raccoon at an "Amateur Radio Club" booth with callsign W6HAM. Claude described this as the easiest result, with the raccoon "not really hiding."
Google Nano Banana Pro: Produced the worst result across all models tested, with an oversized raccoon in the center surrounded by an "ugly white border."
gpt-image-2 (standard quality): Generated a complex scene but did not appear to include the requested raccoon.
gpt-image-2 (high quality, 3840x2160): Successfully generated a detailed Where's Waldo-style scene with the raccoon holding a ham radio visible in the bottom left corner. The output was a 17MB PNG file.
Technical implementation
Willison used a command-line wrapper around the OpenAI Python client:
OPENAI_API_KEY="$(llm keys get openai)" \
uv run https://tools.simonwillison.net/python/openai_image.py \
-m gpt-image-2 \
--quality high --size 3840x2160 \
"[prompt text]"
According to the updated OpenAI image generation cookbook, the model supports multiple output qualities and sizes, with 3840x2160 appearing to be the maximum resolution.
What this means
ChatGPT Images 2.0 demonstrates OpenAI's continued competition with Google's image generation models, particularly Nano Banana 2. The $30 per million token pricing makes high-resolution generation accessible for production use cases. The model's ability to handle complex compositions with both detailed illustrations and readable text represents measurable progress in image generation capability, though Altman's GPT-3 to GPT-5 comparison remains unverified by independent benchmarks.
Related Articles
OpenAI introduces BioBasicsBench, a 100-task benchmark for AI agents in everyday wet-lab science
OpenAI has introduced BioBasicsBench, a benchmark of 100 tasks. It measures how well AI agents interpret raw experimental data and handle everyday wet-lab science.
Salesforce launches Koa Decisions, a Gemma-based model that classifies text, image and audio requests
Salesforce introduced Koa Decisions on November 6, 2026. The decision model is built from Google's open-weight Gemma and accepts text, images and audio. It is the next version of HyperClassifier.
Ant Group releases Ling-3.2-flash: 560B-parameter multimodal model with 1M-token context
Ant Group released Ling-3.2-flash on November 5, a native multimodal model that accepts text, images, and video. It has 560 billion total parameters, activates 25 billion per inference step, and supports a one-million-token context window.
Google releases Gemini Omni 1.2 Flash video model: 15-second clips up to 4K with native audio
Google released Gemini Omni 1.2 Flash on November 5, 2026, a native multimodal video generation and editing model that succeeds Gemini Omni 1.1 Flash. It generates clips up to 15 seconds with native audio at resolutions from 360p to 4K. OpenRouter lists it at $1.50 input and $9 output per 1M with a 131K context window.
Comments
Loading...