NewHow we catch a seller listing the wrong model, and refund you if one does.
kimi logo

How do I use Kimi K3 in Codex CLI?

moonshotai/kimi-k3

Codex CLI reaches Kimi K3 through the Responses API of MartLLM. In ~/.codex/config.toml, add a model provider with base_url set to https://api.martllm.com/v1 and wire_api = "responses", put your MartLLM key in the UBERLLM_API_KEY environment variable, and set model to moonshotai/kimi-k3. model_context_window is set to 1,000,000, the context size of Kimi K3, because Codex assumes 272,000 tokens for models outside the list of OpenAI models it knows.

toml
model = "moonshotai/kimi-k3"
model_provider = "uberllm"
model_context_window = 1000000
web_search = "disabled"

[model_providers.uberllm]
name = "MartLLM"
base_url = "https://api.martllm.com/v1"
env_key = "UBERLLM_API_KEY"
wire_api = "responses"

Codex CLI calls: OpenAI Responses API https://api.martllm.com/v1/responses · Full Codex CLI guide

Replace ull_your_key with your MartLLM API key. One key works in every tool and for every model.

By API, Kimi K3 answers on three addresses with the same MartLLM key and the model id moonshotai/kimi-k3: OpenAI chat completions at https://api.martllm.com/v1/chat/completions, the OpenAI Responses API at https://api.martllm.com/v1/responses and the Anthropic Messages API at https://api.martllm.com/v1/messages.

How much does the Kimi K3 API cost?

On November 20, 2026, Kimi K3 costs $0.0384 per million input tokens and $0.2021 per million output tokens through MartLLM, 98.7% below its list price of $3.00 and $15.00. 118 sellers on the market can serve it. In the 7 days before, half of its 4,728 streamed requests got their first token within 4.9 seconds, with a typical prompt of 100,925 tokens. It reads up to 1,000,000 tokens of context. MartLLM adds no fee to these prices, and a request that fails is not billed.

Prices as of 11:00 UTC.

Kimi K3 in other tools

Other models in Codex CLI