OpenAI's GPT-6 Luna and Google's Gemini 3.5 Flash-Lite are the fastest, cheapest tiers of their families, both with 1M-token context. On list price Luna is lower everywhere, including its long-context tier.
GPT-6 Luna is 3× cheaper on input and 5× cheaper on output at base rates. Even on prompts over 272K tokens, where Luna rises to $0.20/$0.75 per 1M, it remains below Flash-Lite's flat $0.30/$2.50.
| GPT-6 Luna | Gemini 3.5 Flash-Lite | |
|---|---|---|
| Provider | OpenAI | |
| Tier | Budget | Budget |
| Input / 1M tokens | $0.10 | $0.30 |
| Output / 1M tokens | $0.50 | $2.50 |
| Cache hit / 1M | $0.01 | $0.03 |
| Cache write / 1M | $0.125 | — |
| Context window | 1M | 1M |
| Notes | New (Sep 29, 2026 DevDay), fastest & cheapest GPT-6-family tier; >272K tokens: $0.20/$0.75, cache $0.02 | Fastest Flash tier (~350 tok/s); succeeds 3.1 Flash-Lite |
GPT-6 Luna is 3.0× cheaper on input tokens ($0.10 vs $0.30 per 1M). GPT-6 Luna is 5.0× cheaper on output tokens ($0.50 vs $2.50 per 1M).
| Workload | GPT-6 Luna | Gemini 3.5 Flash-Lite | Difference |
|---|---|---|---|
| Support chatbot 10,000 conversations × (2,000 input + 400 output tokens) | $4.00 | $16.00 | GPT-6 Luna saves $12.00 |
| Document Q&A 1,000 questions × (30,000 input + 800 output tokens) | $3.40 | $11.00 | GPT-6 Luna saves $7.60 |
| Coding agent session 4M input tokens (90% cache hits) + 150,000 output tokens | $0.15 | $0.60 | GPT-6 Luna saves $0.45 |
Standard rates per 1M tokens from data on this site. The coding-agent row assumes 90% of input is read from the prompt cache (full input price where a model has no cache rate). Cache-write fees, batch discounts and long-context surcharges listed in the notes are not included. Updated November 2, 2026.
you want the lowest metered cost at any prompt length, with cache hits at $0.01 per 1M and a 30-minute cache included at the base write price.
you need a free tier for prototyping, you are already on Google Cloud or Vertex billing, or Flash-Lite's listed ~350 tokens/s throughput matters more than per-token cost.
Luna's rate steps up above 272K input tokens to $0.20 input, $0.75 output and $0.02 cache hits per 1M. Flash-Lite has no long-context tier, but its flat $0.30 input and $2.50 output are still higher than Luna's long-context rate. There is no prompt length at which Flash-Lite is cheaper on list price.
Luna's cache hit is $0.01 per 1M versus $0.03 for Flash-Lite, and OpenAI's $0.125 write price covers a 30-minute cache. Google bills an hourly storage fee for explicit caches, which adds cost for caches kept warm all day.
Google's free tier includes Flash-Lite input and output, which makes it a sensible starting point for hobby projects or internal tools with tiny volume. For production volume, both providers' batch modes halve prices, so the ratio between them stays about the same.
Yes, on input, output and cache hits, at both short and long prompt lengths.
Google's pricing page lists a single rate for Flash-Lite with no higher tier for long prompts.
Flash-Lite is listed at about 350 tokens per second as Google's fastest Flash tier; Luna is OpenAI's fastest GPT-6-family tier. Measure both on your own prompts.
How we choose: prices come from each vendor's official pricing page and are re-checked every week. We take no affiliate fees, so rankings aren't for sale.