Compare / GPT-5.4 nano vs GPT-6 Luna

GPT-5.4 nano vs GPT-6 Luna

Which is cheaper, which is more capable, and which does better at each kind of work, from 18 results on 4 sources that measured both models.

OpenAIGPT-5.4 nano Released 17 Mar 2026 · $0.20 in · $1.25 out
OpenAIGPT-6 Luna Released 22 Sep 2026 · $0.10 in · $0.50 out
GPT-5.4 nano wins 11 too close to callGPT-6 Luna wins 16

Of the 18 results both models have, the better value on each. Results whose reported ranges overlap are too close to call.

Results as of 19 November 2026 (catalogue release 2026-11-19-3608133e8331).

GPT-5.4 nano or GPT-6 Luna? The short answer

Pick GPT-6 Luna: it scores higher (Intelligence Index 38.1 against 20.7) and costs less.

  • GPT-6 Luna is cheaper to run: $0.002 against $0.005 per game of bullet chess on BulletBench (3.1 times less).
  • GPT-6 Luna scores higher on the Artificial Analysis Intelligence Index: 38.1 against 20.7.
  • GPT-6 Luna is newer: released 22 Sep 2026, against 17 Mar 2026 for GPT-5.4 nano.
  • On third-party benchmarks, on the results both models have: GPT-6 Luna does better for agents and tool use, reasoning and knowledge, writing and sticking to the facts.

Head to head

20.7
Intelligence Index
38.1
$0.46
Price per million tokens, blendedlower is better
$0.20
400,000
Context window, tokens
1,050,000

Every fact side by side

GPT-5.4 nanoGPT-6 Luna
Price per million tokens$0.20 in · $1.25 out$0.10 in · $0.50 out
Blended (3 in : 1 out)$0.46$0.20
Cheapest host, blended$0.46 (Azure)$0.20 (Amazon Bedrock)
Cost per game of bullet chess (BulletBench)$0.005$0.002
Cached input, per million$0.020 cached input$0.010 cached input
Intelligence Index20.738.1
Spring Prompt overall–38
Output speed, tokens a second–127
Context window400,000 tokens1,050,000 tokens
Released17 Mar 202622 Sep 2026 (newer)
WeightsClosedClosed
Developer based inthe USthe US

Prices are the developer's list price where we have read it, otherwise the typical host on OpenRouter; see every model's price. Intelligence and speed from Artificial Analysis.

Which is better for what

Each model's rank among the models on sale today, for each kind of work, from our own benchmarks and licensed sources. Further right is better; the better rank is in green.

Product listingsTurning a sparse product feed and photos into listings that can go live not measured 10 of 35
Decks from an analysisTurning a finished analysis into a deck you could present as it is not measured 12 of 35
User surveysPlanning a user survey and reading its results without being misled not measured 14 of 26
Marketing planningPlanning a year of ad spend without overspending not measured 7 of 35
Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls 74 of 159 53 of 159
Professional workReal tasks from banking, consulting and law, business documents and freelance projects not measured 25 of 51
Reasoning and knowledgeHard questions across science, maths and general knowledge 96 of 187 48 of 187
WritingWhat people prefer in blind comparisons, and judged writing quality 103 of 169 61 of 169
Sticking to the factsSummarising without inventing things, and factual answers 82 of 139 89 of 139
Speed under pressureGood decisions against a real clock (fast chess) 33 of 38 28 of 38

From our research

Every shared result

Each model's best result on every measure both have, grouped by benchmark, with the setting that produced it. Under each measure, every model that benchmark has measured is a grey tick, best to the right, so you can see where the two sit in the field. The headline measures of each benchmark come first; the rest are under "Show all". Green marks the better value; where a source reports ranges that overlap, neither is marked.

MeasureGPT-5.4 nanoGPT-6 Luna
BulletBench measured by Spring Prompt · 3 results
Cost per game · Bullet 60sUS dollars, lower is better 34 · 21 of 62 measured $0.0050no reasoning $0.0016no reasoning
Ladder Elo · Bullet 60sladder Elo, higher is better 59 · 52 of 62 measured 33no reasoning 249no reasoning
Games lost on time · Bullet 60s% of games, lower is better 51 · 39 of 62 measured 66.7%no reasoning 50.0%no reasoning
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 3 results
Overall · TextArena rating, higher is better 137 · 78 of 387 measured 1,401high reasoning 1,443max reasoning
Overall · Text factualityArena rating, higher is better 117 · 98 of 163 measured 1,418high reasoning 1,430max reasoning
Business, management and finance · TextArena rating, higher is better 132 · 71 of 380 measured 1,405high reasoning 1,447max reasoning
Artificial Analysis reported by Artificial Analysis · 3 results
Artificial Analysis Intelligence Indexindex score, higher is better 107 · 39 of 249 measured 20.7xhigh reasoning 38.1max reasoning
Humanity's Last Exam% of questions, higher is better 89 · 53 of 250 measured 28.3%xhigh reasoning 38.5%max reasoning
Terminal-Bench 4.0% of tasks, higher is better 65 · 37 of 120 measured 0.5%xhigh reasoning 12.6%max reasoning
SimpleQA Verified (Epoch AI) reported by SimpleQA Verified (Epoch AI) · 1 result
Correct answers% of questions, higher is better 72 · 40 of 76 measured 11.7%high reasoning 41.4%max reasoning
Show all 18 results (8 more)
MeasureGPT-5.4 nanoGPT-6 Luna
BulletBench measured by Spring Prompt · 2 results
Invalid moves · Bullet 60s% of moves, lower is better 61 · 1 of 62 measured 3.1%no reasoning 0.0%no reasoning
Median move time · Bullet 60smilliseconds, lower is better 33 · 39 of 62 measured 1.1 sno reasoning 1.5 sno reasoning
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 4 results
Creative writing · TextArena rating, higher is better 170 · 77 of 385 measured 1,336high reasoning 1,413max reasoning
Expert prompts · TextArena rating, higher is better 117 · 40 of 338 measured 1,437high reasoning 1,498max reasoning
Instruction following · TextArena rating, higher is better 139 · 57 of 387 measured 1,387high reasoning 1,448max reasoning
Writing, literature and language · TextArena rating, higher is better 156 · 79 of 386 measured 1,363high reasoning 1,421max reasoning
Artificial Analysis reported by Artificial Analysis · 2 results
Long-context reasoning (AA-LCR)% of questions, higher is better 78 · 13 of 241 measured 76.7%xhigh reasoning 83.3%max reasoning
SciCode% of problems, higher is better 65 · 31 of 121 measured 47.2%xhigh reasoning 54.6%max reasoning

Quick answers

Is GPT-5.4 nano better than GPT-6 Luna?

Pick GPT-6 Luna: it scores higher (Intelligence Index 38.1 against 20.7) and costs less.

Which is cheaper, GPT-5.4 nano or GPT-6 Luna?

GPT-6 Luna is cheaper to run: $0.002 against $0.005 per game of bullet chess on BulletBench (3.1 times less).

Which is better for agents and tool use, GPT-5.4 nano or GPT-6 Luna?

GPT-6 Luna ranks 53 of 159 models on sale for multi-step tasks with tools: support desks, coding agents, function calls, against 74 for GPT-5.4 nano.

Which is better for reasoning and knowledge, GPT-5.4 nano or GPT-6 Luna?

GPT-6 Luna ranks 48 of 187 models on sale for hard questions across science, maths and general knowledge, against 96 for GPT-5.4 nano.

Which is better for writing, GPT-5.4 nano or GPT-6 Luna?

GPT-6 Luna ranks 61 of 169 models on sale for what people prefer in blind comparisons, and judged writing quality, against 103 for GPT-5.4 nano.

Which is better for sticking to the facts, GPT-5.4 nano or GPT-6 Luna?

GPT-5.4 nano ranks 82 of 139 models on sale for summarising without inventing things, and factual answers, against 89 for GPT-6 Luna.

Which is better for speed under pressure, GPT-5.4 nano or GPT-6 Luna?

GPT-6 Luna ranks 28 of 38 models on sale for good decisions against a real clock (fast chess), against 33 for GPT-5.4 nano.

Which has the bigger context window, GPT-5.4 nano or GPT-6 Luna?

GPT-6 Luna: 1,050,000 tokens, against 400,000 for GPT-5.4 nano.

Run GPT-5.4 nano and GPT-6 Luna on your own prompt.

Benchmarks aren't your task. Test both side by side on your prompt, with every answer checked against your rules and the cost of each one.