MoonshotLLM
Kimi K3 API
moonshotai/kimi-k3Kimi K3 for long-context and tool-using tasks through the Hermes chat API.
- 1.05M context
- 工具
- 视觉
- 推理
在 Playground 中运行 Kimi K3
Send a chat request with your account and see the exact token charge. Code needs an API key.
参数
请求预览
{"model": "moonshotai/kimi-k3","messages": [],"stream": true,"max_tokens": 1024}网页 Playground 使用你的登录会话发送此请求体。在代码中请使用 API Key 发送相同内容。
结果
发送一条消息,用你的账号运行 Kimi K3。
收到回复后,这里会显示用量和扣费。
Kimi K3 价格
美元 / 百万 token。按每次请求的提示词长度选择档位。
- Reserve请求被接受时,Hermes 会按估算的提示词长度加上 max_tokens(未设置时为 8,192)预留额度。设置 max_tokens 可以减少预留。
- Settle响应结束后,按模型返回的实际用量结算:提示词 token 按输入价、命中缓存的 token 按缓存读取价、输出 token 按输出价,预留的剩余部分立即释放。
- Failures模型拒绝请求时不收费。流式输出在已生成内容后中断时,会按已生成文本估算并扣费。
Kimi K3 能做什么
Moonshot introduces Kimi K3 as a flagship model; its product site highlights long-horizon coding, research and agent tasks. Product features such as hosted agents and creation tools are separate from the Hermes chat endpoint.
The Hermes catalog lists moonshotai/kimi-k3 with a 1,048,576-token context, vision, tools, reasoning and streaming. Structured-output support is not listed for this route. Check the current model status and token pricing before sending requests.
使用场景
Review longer materials
Analyze documents or code while reserving context for the response.
Build tool-using workflows
Handle tool calls in your application and evaluate results for the task.
Reason over images and text
Combine supported image input with written instructions.
用代码调用 Kimi K3
任意 OpenAI SDK 都可用:设置 base URL 和你的 Hermes Key。
- 1
创建 API Key
In the dashboard, with the chat:create scope. API 密钥 - 2
Point your SDK at Hermes
Base URLhttps://api.hermes-ai.net/api/v1 - 3
发送对话补全请求
POST /chat/completionswithmoonshotai/kimi-k3
代码示例
Set HERMES_API_KEY in your environment.
export HERMES_API_KEY=your_api_keycurl https://api.hermes-ai.net/api/v1/chat/completions \
-H "Authorization: Bearer $HERMES_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "moonshotai/kimi-k3",
"messages": [{"role": "user", "content": "Say hello in one sentence."}],
"max_tokens": 200
}'请求
POST https://api.hermes-ai.net/api/v1/chat/completions 带 chat:create 权限的 API Key,或已登录会话。
| 字段 | 类型 | 说明 |
|---|---|---|
| model | string | moonshotai/kimi-k3 |
| messages | array | OpenAI 对话消息:system、user、assistant、tool 角色;text 和 image_url 内容块。 |
| stream | boolean | SSE 流式输出,最后一个事件包含 usage。 |
| max_tokens | integer | 输出上限,同时决定预留额度。 |
| tools / tool_choice | array / object | 仅支持 function 工具。 |
| response_format | object | 模型支持时可用 JSON 模式和 JSON schema。 |
| temperature, top_p, stop, seed, reasoning_effort | 原样透传。 |
响应
| 字段 | 说明 |
|---|---|
| choices[].message / delta | OpenAI chat completion 结构。 |
| usage | 模型返回的 prompt_tokens、completion_tokens 和 prompt_tokens_details.cached_tokens。 |
| hermes.charged_micros | 精确扣费(微美元,JSON 响应)。流式请求可在 GET /api/v1/chat/requests 查看。 |
| x-hermes-request | 请求 id 响应头,账单流水中使用同一 id。 |
| error.code | 文档中的 Hermes 错误码;请按错误码而非消息文本分支。 |
Kimi K3 API 常见问题
Answers built from the model's catalog entry and the Hermes AI API contract.
Is schema-constrained JSON listed for this route?
The Hermes catalog does not list structured outputs for Kimi K3. Validate any JSON returned from a prompt in your application.
What is the Kimi K3 API on Hermes AI?
Kimi K3 by Moonshot is served through the OpenAI-compatible chat completions endpoint, https://api.hermes-ai.net/api/v1/chat/completions, with the model ID moonshotai/kimi-k3.
What is the context window of Kimi K3?
1,048,576 tokens, shared by the prompt and the output.
Does Kimi K3 support vision, tools and JSON output?
Supported: 工具, 视觉 和 推理. Not supported: JSON 和 音频. Streaming works with stream: true.
How is Kimi K3 billed?
Per token, from prepaid credits. When a request is accepted, credits are reserved for the estimated prompt plus max_tokens of output (8,192 when you set none); when it finishes, the exact charge is taken from the usage the model reports and the rest is released. A rejected request is not charged. Rates are in the pricing table.
Can I use the OpenAI SDK with Kimi K3?
Yes. Set the base URL to https://api.hermes-ai.net/api/v1 and use your Hermes API key; send moonshotai/kimi-k3 as the model. Chat completions, streaming, tools and JSON output work unchanged.
What happens when I hit a rate limit?
Too many requests per minute return 429 RATE_LIMITED and too many in flight return 429 CONCURRENCY_LIMITED. If the model provider rate limits, you get 429 UPSTREAM_RATE_LIMITED and nothing is charged. Retry with backoff.
How do I get an API key for Kimi K3?
Create one under API 密钥 in the dashboard with the chat:create scope, then add credits on the Billing page.