Kimi K2 (2025) API: Pricing, Features & Alternatives
Kimi K2 was Moonshot AI's 1-trillion-parameter (32B active) Mixture-of-Experts model released in mid-2025. It has been superseded by Kimi K3, which adds a 1M-token context and frontier reasoning. This page is kept as a reference; use Kimi K3 through BestLLM for current workloads.
Key specifications
| Provider | Moonshot AI |
| Context Window | 128K |
| Max Output | 8K |
| Modalities | Text |
| Release Date | 2025-07 |
Kimi K2 (2025) pricing
Reference pricing for Kimi K2 (2025). Not currently available through BestLLM.
| Tier | Input (per 1M tokens) | Output (per 1M tokens) |
|---|---|---|
| Input | $0.60 | — |
| Output | — | $1.80 |
Features and capabilities
Key capabilities of Kimi K2 (2025):`
- 128K context window
- 1T total / 32B active parameters (MoE)
- Strong agentic and tool-use performance
- Superseded by Kimi K3
API integration example
Direct API access for Kimi K2 (2025):
# Previous generation — see Kimi K3 for the current flagship
curl -X POST https://api.moonshot.cn/v1/chat/completions \
-H "Authorization: Bearer $MOONSHOT_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "kimi-k2",
"messages": [{"role": "user", "content": "Hello"}]
}'Best for
Legacy integrations; new projects should use Kimi K3.
Frequently asked questions
Is Kimi K2 (2025) available through BestLLM?
Kimi K2 (2025) is not currently available through the BestLLM API. This page is kept as a reference with pricing and specs. For production workloads, check our supported models — DeepSeek Chat, DeepSeek Reasoner, Qwen3.8 Max — which are live and accessible through the same OpenAI-compatible API.
What is the pricing for Kimi K2 (2025)?
Kimi K2 (2025) pricing is Input: $0.60 input, Output: / $1.80 output. This reflects direct provider pricing; availability through BestLLM may differ.
What should I use instead of Kimi K2 (2025)?
If you were using Kimi K2 (2025), the closest supported alternative on BestLLM is Kimi K3. It offers a 1M context window and is accessible through the same OpenAI-compatible API.
Need a live model instead?
BestLLM offers DeepSeek, Qwen, Kimi, GLM, and MiniMax through one OpenAI-compatible API with regional routing, spend controls, and transparent per-token billing.