GPT-6 Luna vs MiniMax M3: cost per task
GPT-6 Luna is cheaper on 14 of 14 tasks and about 3.9x cheaper on average (geometric mean). On LMArena Text they are practically tied (1443 vs 1441, within the margin of error).
| Task | GPT-6 Luna | MiniMax M3 | Difference |
|---|---|---|---|
| Answer a chat question | 0.06¢ | 0.12¢ | 2.0x |
| Reply to a customer email | <0.01¢ | 0.07¢ | 9.1x |
| Write a 1,500-word blog post | 0.12¢ | 0.27¢ | 2.3x |
| Write 10 product descriptions | 0.05¢ | 0.21¢ | 3.8x |
| Translate 1,000 words | 0.08¢ | 0.73¢ | 8.8x |
| Proofread one page (500 words) | 0.04¢ | 0.15¢ | 3.4x |
| Summarize a 20-page PDF | 0.16¢ | 0.47¢ | 3.0x |
| Extract data from an invoice | 0.03¢ | 0.12¢ | 3.7x |
| Analyze a 200-row spreadsheet | 0.19¢ | 2.1¢ | 11.0x |
| Answer from 5 retrieved documents (RAG) | 0.05¢ | 0.26¢ | 4.9x |
| Classify 100 support tickets | 0.06¢ | 0.27¢ | 4.3x |
| Review 500 lines of code | 0.31¢ | 0.45¢ | 1.5x |
| Write a SQL query | 0.04¢ | 0.09¢ | 2.3x |
| Run a 50-step coding agent | 4.1¢ | 18¢ | 4.5x |
Specs and list prices
| GPT-6 Luna | MiniMax M3 | |
|---|---|---|
| Input per 1M | $0.10 | $0.30 |
| Output per 1M | $0.50 | $1.20 |
| Cached input per 1M | $0.01 | $0.06 |
| Context | 1.05M | 1M |
| Providers | 7 | 14 |
| Reasoning | default | none |
| LMArena Text | 1443 | 1441 |
| Released | Sep 22, 2026 | May 31, 2026 |
FAQ
Is GPT-6 Luna or MiniMax M3 cheaper?
GPT-6 Luna is cheaper on 14 of 14 tasks and about 3.9x cheaper on average (geometric mean). On LMArena Text they are practically tied (1443 vs 1441, within the margin of error).
How much does GPT-6 Luna cost compared to MiniMax M3 per token?
GPT-6 Luna: $0.10 input and $0.50 output per 1M tokens. MiniMax M3: $0.30 input and $1.20 output. Token prices alone mislead - models use different numbers of tokens for the same task, which is what the task table shows.