GPT-5.6 Luna
Active Parameters
-
Context Length
1.05M
Modality
Multimodal
Architecture
Mixture of Experts (MoE)
License
-
Release Date
9 Jul 2026
Knowledge Cutoff
Feb 2026
Evaluation Benchmarks
No evaluation benchmarks for GPT-5.6 Luna available.
Rankings
Overall Rank
-
Coding Rank
-
About GPT-5.6 Luna
OpenAI's GPT-5.6 Luna is a lightweight, ultrafast, and cost-effective tier designed for high-volume inference, subagent execution, and low-latency workflows.
Technical Specifications
Attention
Attention Structure
Grouped-Query Attention
Attention Heads
-
Key-Value Heads
-
Attention Head Dimension
-
Position Embedding
ROPE
RoPE Theta
-
Sliding Window Attention
-
Sliding Window Size
-
Sliding Window Ratio
-
Linear Attention
-
Linear Attention Ratio
-
Normalization
-
Activation Function
-
Dimensions
Hidden Dimension Size
-
Number of Layers
-
FFN Intermediate Size (Dense)
-
Multi-Token Prediction Heads
-
Tokenizer
Vocabulary Size
-
Mixture of Experts
Total Expert Parameters
-
Number of Experts
-
Active Experts
-
Shared Experts
-
FFN Intermediate Size (per Expert)
-
Dense Layers Before MoE
-
Resources
About GPT-5.6
OpenAI's GPT-5.6 series represents the next-generation frontier intelligence family structured into three tiered variants: Sol (flagship reasoning and autonomous agents), Terra (balanced everyday production workhorse), and Luna (lightweight, ultrafast, cost-effective inference). It introduces new max and ultra (sub-agent coordination) reasoning modes, native programmatic tool calling, a 1M token context window with 128k output tokens, and a February 16, 2026 knowledge cutoff.