GPT-5.6 Luna

Budget

GPT-5.6 Luna is the fastest and cheapest model in the GPT-5.6 family, previewed June 26, 2026. It launched at $1 per million input tokens and $6 per million output, and OpenAI cut it 80 percent on July 30 to $0.20 input, $0.02 cached, and $1.20 output, with a 1 million token context window. Like its siblings, Luna entered as a limited preview to roughly 20 pre-approved organizations under a US Government arrangement, and reached general availability on July 9, 2026. OpenAI is positioning it for high-volume, latency-sensitive workloads where Sol-level reasoning is not required. Its successor GPT-6 Luna shipped on September 22, 2026 at $0.10/$0.50.

Input Price

$0.20

per 1M tokens

Output Price

$1.20

per 1M tokens

Context Window

tokens

Released

2026-07

API access

Capabilities

textvisiontool-usecodereasoning

Key Strengths

✓Lowest pricing in the GPT-5.6 family ($0.20/$1.20 per 1M)
✓1M token context window
✓Frontier-family lineage at budget-tier economics
✓Reasoning, vision, tool use, and code included

Best For

▸High-volume chat and classification
▸Cost-sensitive coding assistants
▸Real-time tool-augmented workflows
▸Latency-sensitive customer-facing applications

Benchmark Scores

Benchmark	Score	Description
SWE-bench	93.0	Real-world software engineering tasks from GitHub issues (SWE-bench Verified)
MMLU-Pro	86.0	General knowledge and reasoning across 57 subjects
GPQA Diamond	92.3	Graduate-level science questions verified by domain experts
OSWorld 2.0	45.6	Computer use across real desktop applications and multi-step GUI tasks
BrowseComp	83.3	Agentic web search and browsing over hard-to-find facts
FrontierCode v1.1	39.8	Agentic coding on frontier software engineering tasks (Main split)
Terminal-Bench 4.0	4.5	Long-horizon agentic work in a terminal across software, science, ML, operations, hardware, security, and media (66 tasks, all-or-nothing verifiers)