GPT-6 Luna costs less: $2.25 less per 1M input tokens and $11.56 less per 1M output tokens. Both take images, call tools and reason before answering. Only GPT-6 Luna can return JSON. Claude Sonnet 5.5 reads up to 1M tokens at once, GPT-6 Luna 1.05M. Both run on one withConflux key.
| Claude Sonnet 5.5 | GPT-6 Luna | |
|---|---|---|
| Input, per 1M tokens | $2.50 | $0.25 |
| Output, per 1M tokens | $12.50 | $0.94 |
| Context | 1M tokens | 1.05M tokens |
| Can use | vision, tools, reasoning | vision, tools, JSON, reasoning |
| Typical answer, our gateway | Not measured yet | Not measured yet |
| First token | Not measured yet | Not measured yet |
Green = better on that row. Speed for every model: API speed.
At your price on withConflux. Change the numbers in the calculator below to match your traffic.
| Workload | Claude Sonnet 5.5 | GPT-6 Luna |
|---|---|---|
| Support chatbot 30,000 messages a month, 1,500 tokens in and 300 out each | $225.00 | $19.69 |
| Coding agent 2,000 tasks a month, 40,000 tokens in and 4,000 out each | $300.00 | $27.50 |
| Document summaries 10,000 documents a month, 8,000 tokens in and 600 out each | $275.00 | $25.63 |
from openai import OpenAI client = OpenAI(base_url="https://withconflux.com/v1", api_key="cflx_live_...") client.chat.completions.create(model="gpt-6-luna", messages=[...]) # or "claude-sonnet-5-5"
Sign up free and call Claude Sonnet 5.5 and GPT-6 Luna with the same key. Each call shows its cost per key.
GPT-6 Luna is cheaper for both input and output.
On withConflux, yes: both are behind the same OpenAI-compatible endpoint. You change the model name.
No. One withConflux key calls both. With your own keys (BYOK) you can use your accounts instead, for €19 a month.
Prices are read from our live model list. Speed is recalculated every hour from the last 7 days. The date is on the page.