GPT-5.6 Luna
openai/gpt-5.6-luna- OpenAI
- Family · GPT-5 Luna
- Released 9 Jul 2026
- File · Image · Text → Text
- Reasoning · adjustable
- Tools · JSON · Vision
Spec & list price
Context1M
Max output128K
Input / 1M$0.20
Output / 1M$1.206× input
Scores · standing
Intelligence37.3#47/489
Coding71.4#23/196
Usage
Tokens / day572.5B▲ 138% 7d
Usage rank#9
Requests / day12.2M
Tool calls / req0.27
Per request · 30 days
Avg input13.0K0.9× median
Avg output3980.6× median
Cache hit66%about median
Reasoning share42%0.7× median
Tool-call failures2.4%1.3× median
Consumption
Peak 1.3T 28 SepLow 177.7B 18 OctRank today #9
News & trends
7 mentions this week ▼ 50%this model · videos, posts, articles · 30 days
Trending · 14 days
- Guys, they are switching GPT-6 Luna with GPT-5.6 Luna for free users (its a celebration if you understand)76 upvotes
- สมาร์ทวอทช์ ที่มาพร้อม GPT 5.6 LUNA ตอบไว ประมวลผลเร็ว #kieslectelite2 #smartwatch #kieslect8.9K views
- @ArtificialAnlys on X51.9K views
- V7 cuts costs 78% while boosting accuracy with GPT‑5.6 Luna27.6K views
News · latest from OpenAI
- GPT-6.1 Terra vs. Claude Opus 5.5: Cheaper and faster in three tests at maxthenewstack.io · 2 Nov
- Chinese labs now target OpenAI's GPT-6 Astra for distillationfork.zone · 1 Nov
- GPT-6.1 Terrasimonwillison.net · 31 Oct
- GPT-5.6 Sol can beat Factorio on peaceful moderyanmadden.net · 31 Oct
Benchmarks
bar = share of the top score · Benchmark data: Artificial AnalysisSpeed · median across 8 providers, latest day
Output speed#77/334100 tok/s1.5× faster than median
Fastest provider138 tok/sbest of 8 providers
First token#1961.29 s1.3× longer than median
#5 cheapest of 14 with a similar score
Value
Intelligence Index
Blended price per 1M tokens (log)
Technical details
- Reasoning
- on · medium
- Tokenizer
- GPT
- Knowledge cutoff
- 2026-02-16
- Moderation
- moderated
Reasoning effort
- max
- xhigh
- high
- medium · default
- low
- none
Request parameters 11
- tools
- tool_choice
- response_format
- structured_outputs
- reasoning
- reasoning_effort
- include_reasoning
- max_tokens
- max_completion_tokens
- seed
+1 more
- verbosity
Interesting facts
- Less output room than similar models (128.0K against a 235.9K median).
- Weekdays average 420.6B tokens a day against 449.3B on weekends.
When to use
Best for
- Cost-aware coding and data analysis
- Reasoning with adjustable effort
- Very long document question answering
- Image and file understanding in apps
- Tool-calling assistants at steady volume
Think twice for
- Long autonomous agent runs
- Self-hosting (closed weights)
- Image generation
Research papers
- ExploreRepair: Evidence-Specialized Agents for Vulnerability Repair under Information Constraints29 Oct
- How Deterministic SAST and AI Security Agents Divide the Work: Coverage, Harness Policy and Cost in Snyk VulnBench 2.028 Oct
- CiteAnchor: Source-Aware Grounding for Hallucinated Citation Detection and Resolution21 Oct
- TraceLite: Efficient Trace Reduction for SWE Agents via Optical Compression and Redundancy Removal20 Oct
- Zephyrus-R1: Agentic Reinforcement Learning for Weather Reporting19 Oct
- SkillNEO: Non-Elitist Evolutionary Optimization of Agent Skills14 Oct
- Evaluating Large Language Models for Multilingual Abstractive Summarization: A Comprehensive Study of English and Bengali13 Oct
- Looking Complete Is Not Being Grounded: Evaluating LLM Sitreps Against Human-Authored Situation Reports13 Oct
Same family
GPT-5 LunaAlternatives
other vendors · similar capability- 37.9
- 34.3
- 33.7
- C33.9
Using GPT-5.6 Luna at volume?
Sign in to ask for a volume discount. You will continue in the OSMEXO dashboard.











