GPT-5.6 Luna vs Kimi K3
Price, context and benchmark scores side by side — every number from the live model catalogue.
| GPT-5.6 Lunaopenai | Kimi K3moonshotai | |
|---|---|---|
| Input / 1M tokens | $0.500 | $3.00 |
| Output / 1M tokens | $3.00 | $15.00 |
| 1M in + 300K out | $1.40 | $7.50 |
| Context window | 1.1M | 1.0M |
| Intelligence index | 51.2 | 57.1 |
| Coding index | 71.4 | 76.2 |
| Agentic index | 45.6 | 50.1 |
| Cheapest provider | OpenAI | BaseTen |
| Providers serving it | 2 | 6 |
| Weights | Closed | Open |
| Image input | Yes | Yes |
| Tool calling | Yes | Yes |
| Reasoning mode | Yes | Yes |
| Released | Jul 9, 2026 | Jul 16, 2026 |
Which one to pick
Pick GPT-5.6 Luna if…
- Token cost matters: $3.00 per 1M output vs $15.00.
- You need the bigger context window (1.1M).
Pick Kimi K3 if…
- You want the higher general-intelligence score (57.1).
- Code quality is the priority — it leads on the coding index (76.2).
- You are running agent loops — higher agentic index (50.1).
- You want the option to self-host — weights are public.
- You want provider redundancy: 6 providers serve it.
FAQ
Is GPT-5.6 Luna better than Kimi K3?
Kimi K3 scores higher on general intelligence (51.2 vs 57.1), while GPT-5.6 Luna is 5.0× cheaper per output token ($3.00 vs $15.00). For code, Kimi K3 leads (76.2 coding index). GPT-5.6 Luna takes the larger context at 1.1M.
Which is cheaper, GPT-5.6 Luna or Kimi K3?
GPT-5.6 Luna, at $3.00 per 1M output tokens versus $15.00. On a 1M input + 300K output workload that is $1.40 against $7.50.
Can I self-host GPT-5.6 Luna or Kimi K3?
GPT-5.6 Luna: closed weights, API only. Kimi K3: open weights (moonshotai/Kimi-K3).