Model comparison
Same task, different trade-offs.
We are not looking for an absolute winner. The comparison shows which model better suits your work, budget and required speed.
Models being compared
GPT-5.6 Luna
Older profile · current status unverified
Data from July 31, 2026
Release date unverified
The most economical variant in the GPT-5.6 family for classification, extraction and other repetitive tasks where cost and speed outweigh maximum quality.
Gemini 3.8 Flash
Data from September 27, 2026
Published September 2, 2026
Processes text, images, audio and video; the output is text.
Price, speed and usage
Low · standard rates from July 30, 2026: $0.20 per million input tokens and $1.20 per million output tokens
$0.75 input / $3.75 output per million tokens through December 31, 2026. From January 1, 2027: $1.50 / $7.50.
Fast; the paid Fast mode costs twice as much
We have not independently measured speed on a comparable basis.
OpenAI API and the Codex tool
Gemini API and Google AI Studio
Varies by product and mode
1,048,576 input tokens; output capped at 65,536 tokens.
Text and, depending on the service used, image inputs
Text, images, video, audio and PDF → text
GPT-5.6 Luna
Use when…
- Classifying and labeling large volumes of text
- An initial low-cost pass before a stronger model
- Simpler automation with precisely controllable output
Choose another model when…
- The hardest analyses and strategic decisions
- Complex programming without subsequent review
Gemini 3.8 Flash
Use when…
- Analysis of documents, audio and video
- Programming and multi-step business tasks
Choose another model when…
- A different model is needed to generate images or audio directly.
Original sources