HIGH-VOLUME TIERGPT-6 FAMILY1.05M CONTEXTHALF THE 5.6 LUNA PRICETEXT + VISION
GPT-6 Luna API Pricing
GPT-6 Luna is OpenAI's high-volume tier, released September 22, 2026 at $0.10/M input, $0.01/M cached input and $0.50/M output — half the GPT-5.6 Luna card of $0.20 / $1.20, on the same 1.05M context window as the rest of the GPT-6 family. Prompts over 272K input tokens bill on the long-context card at $0.20/M and $0.75/M.
Input - per 1M tokens
$0.10/M
Short context up to 272K input standard
Output - per 1M tokens
$0.50/M
Long context output is $0.75/M standard
Cached input - 90% off
$0.01/M
Prompt cache read price -90%
Effective - agentic blend
$0.0682/M
92/8 split - 82% cache
§ 01 / TERMINAL
GPT-6 Luna API cost calculator.
Live calculator pre-loaded with GPT-6 Luna's standard short-context rates. This is the tier to route bulk work to: at $0.10/M input it is a twentieth of GPT-6.1 Sol, and the cache read costs $0.01/M. Reasoning effort runs from none to max and bills as output.
$ /mo
Workload split
Prompt cache hit rate
Tokens you can process
—
Words equivalent (English)
—
Effective rate
—
§ 02 / SCENARIOS
GPT-6 Luna cost on real workloads.
CODING AGENT
Repo-wide architecture migration
$0.00968/task
RESEARCH
Multi-step diligence brief
$0.00384/brief
LONG DOC ANALYSIS
Reading 150-page contracts
$0.011/doc
ASSISTANT
Executive workflow turn
$0.00031/turn
Exact · o200k_base Auto-counts as you type
Counted with the o200k_base BPE encoding — the same tokenizer the API uses — entirely in your browser. The encoder loads on your first keystroke (~1 MB, one time); your text never leaves this page.
Characters 602
Words 104
Tokens (exact) 152 tokens
Cost as input · uncached $0.00002 USD
Cost as output · uncached $0.00008 USD
Cost as cached input $0.000002 USD
| Model | Input /M | Output /M | Effective blended | Context | Best for |
|---|---|---|---|---|---|
| GPT-6 Luna Current | $0.10 cache $0.01 | $0.50 | $0.0682 agentic 92/8 | 1.05M | High-volume and routing work |
| GPT-6.1 Sol | $2.00 cache $0.10 | $10.00 | $1.29 +19× | 1.05M | Coding and long-running agents |
| GPT-6 Astra | $10.00 cache $1.00 | $50.00 | $6.82 +100× | 1.05M | OpenAI's top-of-stack reasoning tier |
| GPT-5.6 Sol | $4.00 cache $0.40 | $20.00 | $2.73 +40× · promo | 1.05M | The Daybreak-programme Sol, still listed |
| Claude Opus 5.5 | $4.00 cache $0.20 | $20.00 | $2.58 +38× | 1M | Anthropic's frontier alternative |
| Grok 4.7 | $2.00 cache $0.50 | $6.00 | $1.19 +17× | 500K | Cheaper frontier coding model |
| Gemini 3.8 Flash | $0.75 cache $0.075 | $3.75 | $0.481 +7.0× · promo | 1M | Multimodal high-volume alternative |
Compare siblings
GPT-6 Luna has published scores on 9 benchmarks we track; its best placing is #39 of 105 on ARC-AGI-2. 4 of these scores come from an independent leaderboard; the rest are the lab's own reports.
| Benchmark | Score | Rank | Leader | Source |
|---|---|---|---|---|
| ARC-AGI-2 Reasoning | 59.3% ARC Prize verified, best result | #39 of 105 | GPT-6 Astra · 95% | Official leaderboard ARC Prize Foundation |
| ARC-AGI-1 Reasoning | 86.7% ARC Prize verified, best result | #42 of 99 | 6 tied · 98.5% | Official leaderboard ARC Prize Foundation |
| DeepSWE Coding | 66.6% max effort | #15 of 27 | Gemini 4 Argon · 77.9% | Vendor-reported OpenAI |
| ARC-AGI-3 Reasoning · Standard harness | 0.2% ARC Prize verified, best result | #18 of 24 | GPT-6 Astra · 62.7% | Official leaderboard ARC Prize Foundation |
| FrontierCode Coding · v1.1 Main (100 hardest tasks) | 42.4% max effort | #10 of 13 | Claude Opus 5.5 · 54.4% | Vendor-reported OpenAI |
| Terminal-Bench 4.0 Agents and computer use | 16.4% max effort, Codex harness, 330 trials, ±2.7 pts (95% CI) | #27 of 34 | Kimi K3.1 Pro · 65.7% | Official leaderboard Terminal-Bench (Stanford, Laude Institute) |
| Agents' Last Exam Agents and computer use · ALE v1 | 50.9% max effort | #4 of 5 | GPT-6 Astra · 59.3% | Vendor-reported OpenAI |
| OSWorld 2 Agents and computer use · 2.0, offline subset, partial score | 52.7% max effort | #6 of 7 | GPT-6 Astra · 73.5% | Vendor-reported OpenAI |
| AutomationBench Agents and computer use · Main set (as reported) | 20.7% max effort, v1.0.6 | #20 of 22 | MiniMax M3.1 Flash · 62.8% | Vendor-reported OpenAI |
GPT-6 Luna API pricing FAQ.
GPT-6 Luna pricing questions - the cheapest current OpenAI card, and where the cheap tier stops being cheap.
Q · 01 How much does GPT-6 Luna cost? +
OpenAI lists
gpt-6-luna at $0.10/M input, $0.01/M cached input, $0.125/M cache writes and $0.50/M output for prompts up to 272K input tokens.Q · 02 Is it cheaper than GPT-5.6 Luna? +
Yes, by half:
$0.10 / $0.50 against $0.20 / $1.20, and the cache read drops from $0.02/M to $0.01/M. Blended at our standard mix that is $0.0682/M against $0.152/M.Q · 03 How does it compare with GPT-6.1 Sol? +
Twenty times cheaper on input and output alike —
$0.10 / $0.50 against $2.00 / $10.00 — on the same 1.05M context window. Route the volume here and the hard work to GPT-6.1 Sol; the arithmetic of that split is on our routing-by-complexity post.Q · 04 When does long-context pricing apply? +
Above
272K input tokens, where the whole request bills at $0.20/M input, $0.02/M cached input and $0.75/M output. On a cheap tier that step matters more in relative terms than it does on Sol.Q · 05 How much cheaper are Batch and Flex? +
Half the standard card:
$0.05/M input and $0.25/M output. Fast mode runs at double standard instead.Q · 06 What is the context window? +
1.05M tokens with 128K max output and a knowledge cutoff of May 18, 2026 — the same window as GPT-6 Sol and GPT-6 Astra. Text and image input, text output.Q · 07 Is this the model ChatGPT's free tier runs? +
No. ChatGPT's free tier runs a GPT-5.6-family model; GPT-6 Sol and Luna started rolling into ChatGPT Work and Codex on launch day rather than into chat. This page prices the API model, which is sold independently of any ChatGPT plan.