应用程序接口
端点
约定
模型名称
模型名称遵循model:tag
格式,其中model
可以具有可选的命名空间,例如example/model
.一些例子是orca-mini:3b-q4_1
和llama3:70b
.该标签是可选的,如果未提供,则默认为latest
.该标签用于标识特定版本。
持续时间
所有持续时间均以纳秒为单位返回。
流式处理响应
某些终端节点将响应作为 JSON 对象流式传输。可以通过提供{"stream": false}
对于这些终端节点。
生成完成
POST /api/generate
使用提供的模型为给定提示生成响应。这是一个流式处理终结点,因此将有一系列响应。最终响应对象将包含来自请求的统计信息和其他数据。
参数
model
:(必需)模型名称prompt
:生成响应的提示suffix
:模型响应后的文本images
:(可选)base64 编码图像列表(对于多模态模型,例如llava
)
高级参数(可选):
format
:返回响应的格式。格式可以是json
或 JSON 架构options
:Modelfile 文档中列出的其他模型参数,例如temperature
system
:系统消息更改为(覆盖Modelfile
)template
:要使用的提示模板(覆盖Modelfile
)stream
:如果false
响应将作为单个响应对象返回,而不是对象流raw
:如果true
不会对提示应用任何格式。您可以选择使用raw
参数(如果要在对 API 的请求中指定完整模板化提示)keep_alive
:控制模型在请求后加载到内存中的时间(默认值:5m
)context
(已弃用):从上一个请求返回的 context 参数/generate
,这可用于保持简短的对话记忆
结构化输出
通过在format
参数。该模型将生成与架构匹配的响应。请参阅下面的结构化输出示例。
JSON 模式
通过设置format
参数设置为json
.这会将响应构建为有效的 JSON 对象。请参阅下面的 JSON 模式示例。
[!重要] 请务必指示模型在
prompt
.否则,模型可能会生成大量空格。
示例
生成请求 (Streaming)
请求
curl http://localhost:11434/api/generate -d '{
"model": "llama3.2",
"prompt": "Why is the sky blue?"
}'
响应
返回 JSON 对象流:
{
"model": "llama3.2",
"created_at": "2023-08-04T08:52:19.385406455-07:00",
"response": "The",
"done": false
}
流中的最终响应还包括有关生成的其他数据:
total_duration
:生成响应所花费的时间load_duration
:加载模型所花费的时间(以纳秒为单位)prompt_eval_count
:提示中的令牌数prompt_eval_duration
:评估提示所花费的时间(以纳秒为单位)eval_count
:响应中的令牌数eval_duration
:生成响应所花费的时间(以纳秒为单位)context
:此响应中使用的对话的编码,可以在下一个请求中发送以保持对话记忆response
:如果响应是流式的,则为空,如果未流式响应,则此响应将包含完整的响应
要计算以每秒令牌数 (token/s) 为单位生成响应的速度,请除以eval_count
/ eval_duration
* 10^9
.
{
"model": "llama3.2",
"created_at": "2023-08-04T19:22:45.499127Z",
"response": "",
"done": true,
"context": [1, 2, 3],
"total_duration": 10706818083,
"load_duration": 6338219291,
"prompt_eval_count": 26,
"prompt_eval_duration": 130079000,
"eval_count": 259,
"eval_duration": 4232710000
}
请求 (无流式处理)
请求
当流式传输关闭时,可以在一个回复中收到响应。
curl http://localhost:11434/api/generate -d '{
"model": "llama3.2",
"prompt": "Why is the sky blue?",
"stream": false
}'
响应
如果stream
设置为false
,响应将是单个 JSON 对象:
{
"model": "llama3.2",
"created_at": "2023-08-04T19:22:45.499127Z",
"response": "The sky is blue because it is the color of the sky.",
"done": true,
"context": [1, 2, 3],
"total_duration": 5043500667,
"load_duration": 5025959,
"prompt_eval_count": 26,
"prompt_eval_duration": 325953000,
"eval_count": 290,
"eval_duration": 4709213000
}
请求(带后缀)
请求
curl http://localhost:11434/api/generate -d '{
"model": "codellama:code",
"prompt": "def compute_gcd(a, b):",
"suffix": " return result",
"options": {
"temperature": 0
},
"stream": false
}'
响应
{
"model": "codellama:code",
"created_at": "2024-07-22T20:47:51.147561Z",
"response": "\n if a == 0:\n return b\n else:\n return compute_gcd(b % a, a)\n\ndef compute_lcm(a, b):\n result = (a * b) / compute_gcd(a, b)\n",
"done": true,
"done_reason": "stop",
"context": [...],
"total_duration": 1162761250,
"load_duration": 6683708,
"prompt_eval_count": 17,
"prompt_eval_duration": 201222000,
"eval_count": 63,
"eval_duration": 953997000
}
请求 (结构化输出)
请求
curl -X POST http://localhost:11434/api/generate -H "Content-Type: application/json" -d '{
"model": "llama3.1:8b",
"prompt": "Ollama is 22 years old and is busy saving the world. Respond using JSON",
"stream": false,
"format": {
"type": "object",
"properties": {
"age": {
"type": "integer"
},
"available": {
"type": "boolean"
}
},
"required": [
"age",
"available"
]
}
}'
响应
{
"model": "llama3.1:8b",
"created_at": "2024-12-06T00:48:09.983619Z",
"response": "{\n \"age\": 22,\n \"available\": true\n}",
"done": true,
"done_reason": "stop",
"context": [1, 2, 3],
"total_duration": 1075509083,
"load_duration": 567678166,
"prompt_eval_count": 28,
"prompt_eval_duration": 236000000,
"eval_count": 16,
"eval_duration": 269000000
}
请求(JSON 模式)
[!重要] 什么时候
format
设置为json
,则输出将始终是格式正确的 JSON 对象。指示模型以 JSON 格式响应也很重要。
请求
curl http://localhost:11434/api/generate -d '{
"model": "llama3.2",
"prompt": "What color is the sky at different times of the day? Respond using JSON",
"format": "json",
"stream": false
}'
响应
{
"model": "llama3.2",
"created_at": "2023-11-09T21:07:55.186497Z",
"response": "{\n\"morning\": {\n\"color\": \"blue\"\n},\n\"noon\": {\n\"color\": \"blue-gray\"\n},\n\"afternoon\": {\n\"color\": \"warm gray\"\n},\n\"evening\": {\n\"color\": \"orange\"\n}\n}\n",
"done": true,
"context": [1, 2, 3],
"total_duration": 4648158584,
"load_duration": 4071084,
"prompt_eval_count": 36,
"prompt_eval_duration": 439038000,
"eval_count": 180,
"eval_duration": 4196918000
}
的值response
将是包含 JSON 的字符串,类似于:
{
"morning": {
"color": "blue"
},
"noon": {
"color": "blue-gray"
},
"afternoon": {
"color": "warm gray"
},
"evening": {
"color": "orange"
}
}
请求(带图片)
要将图像提交到多模态模型,例如llava
或bakllava
,提供 base64 编码的列表images
:
请求
curl http://localhost:11434/api/generate -d '{
"model": "llava",
"prompt":"What is in this picture?",
"stream": false,
"images": ["******"]
}'
响应
{
"model": "llava",
"created_at": "2023-11-03T15:36:02.583064Z",
"response": "A happy cartoon character, which is cute and cheerful.",
"done": true,
"context": [1, 2, 3],
"total_duration": 2938432250,
"load_duration": 2559292,
"prompt_eval_count": 1,
"prompt_eval_duration": 2195557000,
"eval_count": 44,
"eval_duration": 736432000
}
请求 (Raw 模式)
在某些情况下,您可能希望绕过模板系统并提供完整提示。在这种情况下,您可以使用raw
参数来禁用模板。另请注意,raw 模式不会返回上下文。
请求
curl http://localhost:11434/api/generate -d '{
"model": "mistral",
"prompt": "[INST] why is the sky blue? [/INST]",
"raw": true,
"stream": false
}'
请求(可重现的输出)
对于可重现的输出,请将seed
转换为数字:
请求
curl http://localhost:11434/api/generate -d '{
"model": "mistral",
"prompt": "Why is the sky blue?",
"options": {
"seed": 123
}
}'
响应
{
"model": "mistral",
"created_at": "2023-11-03T15:36:02.583064Z",
"response": " The sky appears blue because of a phenomenon called Rayleigh scattering.",
"done": true,
"total_duration": 8493852375,
"load_duration": 6589624375,
"prompt_eval_count": 14,
"prompt_eval_duration": 119039000,
"eval_count": 110,
"eval_duration": 1779061000
}
生成请求(带选项)
如果要在运行时而不是在 Modelfile 中为模型设置自定义选项,可以使用options
参数。此示例设置所有可用选项,但您可以单独设置其中任何一个选项,并省略您不想覆盖的选项。
请求
curl http://localhost:11434/api/generate -d '{
"model": "llama3.2",
"prompt": "Why is the sky blue?",
"stream": false,
"options": {
"num_keep": 5,
"seed": 42,
"num_predict": 100,
"top_k": 20,
"top_p": 0.9,
"min_p": 0.0,
"typical_p": 0.7,
"repeat_last_n": 33,
"temperature": 0.8,
"repeat_penalty": 1.2,
"presence_penalty": 1.5,
"frequency_penalty": 1.0,
"mirostat": 1,
"mirostat_tau": 0.8,
"mirostat_eta": 0.6,
"penalize_newline": true,
"stop": ["\n", "user:"],
"numa": false,
"num_ctx": 1024,
"num_batch": 2,
"num_gpu": 1,
"main_gpu": 0,
"low_vram": false,
"vocab_only": false,
"use_mmap": true,
"use_mlock": false,
"num_thread": 8
}
}'
响应
{
"model": "llama3.2",
"created_at": "2023-08-04T19:22:45.499127Z",
"response": "The sky is blue because it is the color of the sky.",
"done": true,
"context": [1, 2, 3],
"total_duration": 4935886791,
"load_duration": 534986708,
"prompt_eval_count": 26,
"prompt_eval_duration": 107345000,
"eval_count": 237,
"eval_duration": 4289432000
}
加载模型
如果提供空提示,则模型将被加载到内存中。
请求
curl http://localhost:11434/api/generate -d '{
"model": "llama3.2"
}'
响应
返回单个 JSON 对象:
{
"model": "llama3.2",
"created_at": "2023-12-18T19:52:07.071755Z",
"response": "",
"done": true
}
卸载模型
如果提供了空提示,并且keep_alive
parameter 设置为0
,模型将从内存中卸载。
请求
curl http://localhost:11434/api/generate -d '{
"model": "llama3.2",
"keep_alive": 0
}'
响应
返回单个 JSON 对象:
{
"model": "llama3.2",
"created_at": "2024-09-12T03:54:03.516566Z",
"response": "",
"done": true,
"done_reason": "unload"
}
生成聊天完成
POST /api/chat
使用提供的模型在聊天中生成下一条消息。这是一个流式处理终结点,因此将有一系列响应。可以使用"stream": false
.最终响应对象将包含来自请求的统计信息和其他数据。
参数
model
:(必需)模型名称messages
:聊天的消息,可用于保存聊天记录tools
:JSON 中供模型使用的工具列表(如果支持)
这message
object 具有以下字段:
role
:消息的角色system
,user
,assistant
或tool
content
:消息的内容images
(可选):要包含在消息中的图像列表(对于多模态模型,例如llava
)tool_calls
(可选):模型要使用的 JSON 中的工具列表
高级参数(可选):
format
:返回响应的格式。格式可以是json
或 JSON 架构。options
:Modelfile 文档中列出的其他模型参数,例如temperature
stream
:如果false
响应将作为单个响应对象返回,而不是对象流keep_alive
:控制模型在请求后加载到内存中的时间(默认值:5m
)
结构化输出
通过在format
参数。该模型将生成与架构匹配的响应。请参阅下面的 Chat request (Structured outputs) 示例。
示例
聊天请求 (流式处理)
请求
发送包含流式响应的聊天消息。
curl http://localhost:11434/api/chat -d '{
"model": "llama3.2",
"messages": [
{
"role": "user",
"content": "why is the sky blue?"
}
]
}'
响应
返回 JSON 对象流:
{
"model": "llama3.2",
"created_at": "2023-08-04T08:52:19.385406455-07:00",
"message": {
"role": "assistant",
"content": "The",
"images": null
},
"done": false
}
最终回应:
{
"model": "llama3.2",
"created_at": "2023-08-04T19:22:45.499127Z",
"done": true,
"total_duration": 4883583458,
"load_duration": 1334875,
"prompt_eval_count": 26,
"prompt_eval_duration": 342546000,
"eval_count": 282,
"eval_duration": 4535599000
}
聊天请求 (无流式处理)
请求
curl http://localhost:11434/api/chat -d '{
"model": "llama3.2",
"messages": [
{
"role": "user",
"content": "why is the sky blue?"
}
],
"stream": false
}'
响应
{
"model": "llama3.2",
"created_at": "2023-12-12T14:13:43.416799Z",
"message": {
"role": "assistant",
"content": "Hello! How are you today?"
},
"done": true,
"total_duration": 5191566416,
"load_duration": 2154458,
"prompt_eval_count": 26,
"prompt_eval_duration": 383809000,
"eval_count": 298,
"eval_duration": 4799921000
}
聊天请求(结构化输出)
请求
curl -X POST http://localhost:11434/api/chat -H "Content-Type: application/json" -d '{
"model": "llama3.1",
"messages": [{"role": "user", "content": "Ollama is 22 years old and busy saving the world. Return a JSON object with the age and availability."}],
"stream": false,
"format": {
"type": "object",
"properties": {
"age": {
"type": "integer"
},
"available": {
"type": "boolean"
}
},
"required": [
"age",
"available"
]
},
"options": {
"temperature": 0
}
}'
响应
{
"model": "llama3.1",
"created_at": "2024-12-06T00:46:58.265747Z",
"message": { "role": "assistant", "content": "{\"age\": 22, \"available\": false}" },
"done_reason": "stop",
"done": true,
"total_duration": 2254970291,
"load_duration": 574751416,
"prompt_eval_count": 34,
"prompt_eval_duration": 1502000000,
"eval_count": 12,
"eval_duration": 175000000
}
聊天请求(带历史记录)
发送包含对话历史记录的聊天消息。您可以使用相同的方法,使用多发或思维链提示来开始对话。
请求
curl http://localhost:11434/api/chat -d '{
"model": "llama3.2",
"messages": [
{
"role": "user",
"content": "why is the sky blue?"
},
{
"role": "assistant",
"content": "due to rayleigh scattering."
},
{
"role": "user",
"content": "how is that different than mie scattering?"
}
]
}'
响应
返回 JSON 对象流:
{
"model": "llama3.2",
"created_at": "2023-08-04T08:52:19.385406455-07:00",
"message": {
"role": "assistant",
"content": "The"
},
"done": false
}
最终回应:
{
"model": "llama3.2",
"created_at": "2023-08-04T19:22:45.499127Z",
"done": true,
"total_duration": 8113331500,
"load_duration": 6396458,
"prompt_eval_count": 61,
"prompt_eval_duration": 398801000,
"eval_count": 468,
"eval_duration": 7701267000
}
聊天请求(带图片)
请求
发送带有图像的聊天消息。图像应以数组形式提供,单个图像以 Base64 编码。
curl http://localhost:11434/api/chat -d '{
"model": "llava",
"messages": [
{
"role": "user",
"content": "what is in this image?",
"images": ["******"]
}
]
}'
响应
{
"model": "llava",
"created_at": "2023-12-13T22:42:50.203334Z",
"message": {
"role": "assistant",
"content": " The image features a cute, little pig with an angry facial expression. It's wearing a heart on its shirt and is waving in the air. This scene appears to be part of a drawing or sketching project.",
"images": null
},
"done": true,
"total_duration": 1668506709,
"load_duration": 1986209,
"prompt_eval_count": 26,
"prompt_eval_duration": 359682000,
"eval_count": 83,
"eval_duration": 1303285000
}
聊天请求(可重现的输出)
请求
curl http://localhost:11434/api/chat -d '{
"model": "llama3.2",
"messages": [
{
"role": "user",
"content": "Hello!"
}
],
"options": {
"seed": 101,
"temperature": 0
}
}'
响应
{
"model": "llama3.2",
"created_at": "2023-12-12T14:13:43.416799Z",
"message": {
"role": "assistant",
"content": "Hello! How are you today?"
},
"done": true,
"total_duration": 5191566416,
"load_duration": 2154458,
"prompt_eval_count": 26,
"prompt_eval_duration": 383809000,
"eval_count": 298,
"eval_duration": 4799921000
}
聊天请求(带工具)
请求
curl http://localhost:11434/api/chat -d '{
"model": "llama3.2",
"messages": [
{
"role": "user",
"content": "What is the weather today in Paris?"
}
],
"stream": false,
"tools": [
{
"type": "function",
"function": {
"name": "get_current_weather",
"description": "Get the current weather for a location",
"parameters": {
"type": "object",
"properties": {
"location": {
"type": "string",
"description": "The location to get the weather for, e.g. San Francisco, CA"
},
"format": {
"type": "string",
"description": "The format to return the weather in, e.g. 'celsius' or 'fahrenheit'",
"enum": ["celsius", "fahrenheit"]
}
},
"required": ["location", "format"]
}
}
}
]
}'
响应
{
"model": "llama3.2",
"created_at": "2024-07-22T20:33:28.123648Z",
"message": {
"role": "assistant",
"content": "",
"tool_calls": [
{
"function": {
"name": "get_current_weather",
"arguments": {
"format": "celsius",
"location": "Paris, FR"
}
}
}
]
},
"done_reason": "stop",
"done": true,
"total_duration": 885095291,
"load_duration": 3753500,
"prompt_eval_count": 122,
"prompt_eval_duration": 328493000,
"eval_count": 33,
"eval_duration": 552222000
}
加载模型
如果 messages 数组为空,则模型将被加载到内存中。
请求
curl http://localhost:11434/api/chat -d '{
"model": "llama3.2",
"messages": []
}'
响应
{
"model": "llama3.2",
"created_at":"2024-09-12T21:17:29.110811Z",
"message": {
"role": "assistant",
"content": ""
},
"done_reason": "load",
"done": true
}
卸载模型
如果 messages 数组为空且keep_alive
parameter 设置为0
,模型将从内存中卸载。
请求
curl http://localhost:11434/api/chat -d '{
"model": "llama3.2",
"messages": [],
"keep_alive": 0
}'
响应
返回单个 JSON 对象:
{
"model": "llama3.2",
"created_at":"2024-09-12T21:33:17.547535Z",
"message": {
"role": "assistant",
"content": ""
},
"done_reason": "unload",
"done": true
}
创建模型
POST /api/create
从以下位置创建模型:
- 另一个模型;
- safeTensors 目录;或
- 一个 GGUF 文件。
如果要从 safetensors 目录或 GGUF 文件创建模型,则必须为每个文件创建一个 blob,然后使用与每个 blob 关联的文件名和 SHA256 摘要files
田。
参数
model
:要创建的模型的名称from
:(可选)要从中创建新模型的现有模型的名称files
:(可选)文件名字典到 blob 的 SHA256 摘要,用于创建模型adapters
:(可选)用于 LORA 适配器的 blob 的 SHA256 摘要的文件名字典template
:(可选)模型的提示模板license
:(可选)包含模型的许可证的字符串或字符串列表system
:(可选)包含模型的系统提示符的字符串parameters
:(可选)模型的参数字典(有关参数列表,请参阅 Modelfile)messages
:(可选)用于创建对话的消息对象列表stream
:(可选)如果false
响应将作为单个响应对象返回,而不是对象流quantize
(可选):量化非量化(例如 float16)模型
量化类型
类型 | 推荐 |
---|---|
q2_K | |
q3_K_L | |
q3_K_M | |
q3_K_S | |
q4_0 | |
q4_1 | |
q4_K_M | * |
q4_K_S | |
q5_0 | |
q5_1 | |
q5_K_M | |
q5_K_S | |
q6_K | |
q8_0 | * |
示例
创建新模型
从现有模型创建新模型。
请求
curl http://localhost:11434/api/create -d '{
"model": "mario",
"from": "llama3.2",
"system": "You are Mario from Super Mario Bros."
}'
响应
返回 JSON 对象流:
{"status":"reading model metadata"}
{"status":"creating system layer"}
{"status":"using already created layer sha256:22f7f8ef5f4c791c1b03d7eb414399294764d7cc82c7e94aa81a1feb80a983a2"}
{"status":"using already created layer sha256:8c17c2ebb0ea011be9981cc3922db8ca8fa61e828c5d3f44cb6ae342bf80460b"}
{"status":"using already created layer sha256:7c23fb36d80141c4ab8cdbb61ee4790102ebd2bf7aeff414453177d4f2110e5d"}
{"status":"using already created layer sha256:2e0493f67d0c8c9c68a8aeacdf6a38a2151cb3c4c1d42accf296e19810527988"}
{"status":"using already created layer sha256:2759286baa875dc22de5394b4a925701b1896a7e3f8e53275c36f75a877a82c9"}
{"status":"writing layer sha256:df30045fe90f0d750db82a058109cecd6d4de9c90a3d75b19c09e5f64580bb42"}
{"status":"writing layer sha256:f18a68eb09bf925bb1b669490407c1b1251c5db98dc4d3d81f3088498ea55690"}
{"status":"writing manifest"}
{"status":"success"}
量化模型
量化非量化模型。
请求
curl http://localhost:11434/api/create -d '{
"model": "llama3.1:quantized",
"from": "llama3.1:8b-instruct-fp16",
"quantize": "q4_K_M"
}'
响应
返回 JSON 对象流:
{"status":"quantizing F16 model to Q4_K_M"}
{"status":"creating new layer sha256:667b0c1932bc6ffc593ed1d03f895bf2dc8dc6df21db3042284a6f4416b06a29"}
{"status":"using existing layer sha256:11ce4ee3e170f6adebac9a991c22e22ab3f8530e154ee669954c4bc73061c258"}
{"status":"using existing layer sha256:0ba8f0e314b4264dfd19df045cde9d4c394a52474bf92ed6a3de22a4ca31a177"}
{"status":"using existing layer sha256:56bb8bd477a519ffa694fc449c2413c6f0e1d3b1c88fa7e3c9d88d3ae49d4dcb"}
{"status":"creating new layer sha256:455f34728c9b5dd3376378bfb809ee166c145b0b4c1f1a6feca069055066ef9a"}
{"status":"writing manifest"}
{"status":"success"}
从 GGUF 创建模型
从 GGUF 文件创建模型。这files
参数中应填写您要使用的 GGUF 文件的文件名和 SHA256 摘要。在调用此 API 之前,请使用 /api/blobs/:d igest 将 GGUF 文件推送到服务器。
请求
curl http://localhost:11434/api/create -d '{
"model": "my-gguf-model",
"files": {
"test.gguf": "sha256:432f310a77f4650a88d0fd59ecdd7cebed8d684bafea53cbff0473542964f0c3"
}
}'
响应
返回 JSON 对象流:
{"status":"parsing GGUF"}
{"status":"using existing layer sha256:432f310a77f4650a88d0fd59ecdd7cebed8d684bafea53cbff0473542964f0c3"}
{"status":"writing manifest"}
{"status":"success"}
从 Safetensors 目录创建模型
这files
parameter 应包括 safetensors 模型的文件字典,其中包括每个文件的文件名和 SHA256 摘要。使用 /api/blobs/:d igest 将每个文件推送到服务器,然后再调用此 API。文件将保留在缓存中,直到 Ollama 服务器重新启动。
请求
curl http://localhost:11434/api/create -d '{
"model": "fred",
"files": {
"config.json": "sha256:dd3443e529fb2290423a0c65c2d633e67b419d273f170259e27297219828e389",
"generation_config.json": "sha256:88effbb63300dbbc7390143fbbdd9d9fa50587b37e8bfd16c8c90d4970a74a36",
"special_tokens_map.json": "sha256:b7455f0e8f00539108837bfa586c4fbf424e31f8717819a6798be74bef813d05",
"tokenizer.json": "sha256:bbc1904d35169c542dffbe1f7589a5994ec7426d9e5b609d07bab876f32e97ab",
"tokenizer_config.json": "sha256:24e8a6dc2547164b7002e3125f10b415105644fcf02bf9ad8b674c87b1eaaed6",
"model.safetensors": "sha256:1ff795ff6a07e6a68085d206fb84417da2f083f68391c2843cd2b8ac6df8538f"
}
}'
响应
返回 JSON 对象流:
{"status":"converting model"}
{"status":"creating new layer sha256:05ca5b813af4a53d2c2922933936e398958855c44ee534858fcfd830940618b6"}
{"status":"using autodetected template llama3-instruct"}
{"status":"using existing layer sha256:56bb8bd477a519ffa694fc449c2413c6f0e1d3b1c88fa7e3c9d88d3ae49d4dcb"}
{"status":"writing manifest"}
{"status":"success"}
检查 Blob 是否存在
HEAD /api/blobs/:digest
确保服务器上存在用于创建模型的文件 blob(二进制大对象)。这将检查您的 Ollama 服务器,而不是 ollama.com。
查询参数
digest
:blob 的 SHA256 摘要
示例
请求
curl -I http://localhost:11434/api/blobs/sha256:29fdb92e57cf0827ded04ae6461b5931d01fa595843f55d36f5b275a52087dd2
响应
如果 blob 存在,则返回 200 OK,如果不存在,则返回 404 Not Found。
推送 Blob
POST /api/blobs/:digest
将文件推送到 Ollama 服务器以创建 “blob” (二进制大对象)。
查询参数
digest
:文件的预期 SHA256 摘要
示例
请求
curl -T model.gguf -X POST http://localhost:11434/api/blobs/sha256:29fdb92e57cf0827ded04ae6461b5931d01fa595843f55d36f5b275a52087dd2
响应
如果 blob 创建成功,则返回 201 Created ,如果使用的摘要不是预期的,则返回 400 Bad Request。
列出本地模型
GET /api/tags
列出本地可用的模型。
示例
请求
curl http://localhost:11434/api/tags
响应
将返回单个 JSON 对象。
{
"models": [
{
"name": "codellama:13b",
"modified_at": "2023-11-04T14:56:49.277302595-07:00",
"size": 7365960935,
"digest": "9f438cb9cd581fc025612d27f7c1a6669ff83a8bb0ed86c94fcf4c5440555697",
"details": {
"format": "gguf",
"family": "llama",
"families": null,
"parameter_size": "13B",
"quantization_level": "Q4_0"
}
},
{
"name": "llama3:latest",
"modified_at": "2023-12-07T09:32:18.757212583-08:00",
"size": 3825819519,
"digest": "fe938a131f40e6f6d40083c9f0f430a515233eb2edaa6d72eb85c50d64f2300e",
"details": {
"format": "gguf",
"family": "llama",
"families": null,
"parameter_size": "7B",
"quantization_level": "Q4_0"
}
}
]
}
显示模型信息
POST /api/show
显示有关模型的信息,包括详细信息、模型文件、模板、参数、许可证、系统提示符。
参数
model
:要显示的模型的名称verbose
:(可选)如果设置为true
返回详细响应字段的完整数据
示例
请求
curl http://localhost:11434/api/show -d '{
"model": "llama3.2"
}'
响应
{
"modelfile": "# Modelfile generated by \"ollama show\"\n# To build a new Modelfile based on this one, replace the FROM line with:\n# FROM llava:latest\n\nFROM /Users/matt/.ollama/models/blobs/sha256:200765e1283640ffbd013184bf496e261032fa75b99498a9613be4e94d63ad52\nTEMPLATE \"\"\"{{ .System }}\nUSER: {{ .Prompt }}\nASSISTANT: \"\"\"\nPARAMETER num_ctx 4096\nPARAMETER stop \"\u003c/s\u003e\"\nPARAMETER stop \"USER:\"\nPARAMETER stop \"ASSISTANT:\"",
"parameters": "num_keep 24\nstop \"<|start_header_id|>\"\nstop \"<|end_header_id|>\"\nstop \"<|eot_id|>\"",
"template": "{{ if .System }}<|start_header_id|>system<|end_header_id|>\n\n{{ .System }}<|eot_id|>{{ end }}{{ if .Prompt }}<|start_header_id|>user<|end_header_id|>\n\n{{ .Prompt }}<|eot_id|>{{ end }}<|start_header_id|>assistant<|end_header_id|>\n\n{{ .Response }}<|eot_id|>",
"details": {
"parent_model": "",
"format": "gguf",
"family": "llama",
"families": [
"llama"
],
"parameter_size": "8.0B",
"quantization_level": "Q4_0"
},
"model_info": {
"general.architecture": "llama",
"general.file_type": 2,
"general.parameter_count": 8030261248,
"general.quantization_version": 2,
"llama.attention.head_count": 32,
"llama.attention.head_count_kv": 8,
"llama.attention.layer_norm_rms_epsilon": 0.00001,
"llama.block_count": 32,
"llama.context_length": 8192,
"llama.embedding_length": 4096,
"llama.feed_forward_length": 14336,
"llama.rope.dimension_count": 128,
"llama.rope.freq_base": 500000,
"llama.vocab_size": 128256,
"tokenizer.ggml.bos_token_id": 128000,
"tokenizer.ggml.eos_token_id": 128009,
"tokenizer.ggml.merges": [], // populates if `verbose=true`
"tokenizer.ggml.model": "gpt2",
"tokenizer.ggml.pre": "llama-bpe",
"tokenizer.ggml.token_type": [], // populates if `verbose=true`
"tokenizer.ggml.tokens": [] // populates if `verbose=true`
}
}
复制模型
POST /api/copy
复制模型。从现有模型创建具有其他名称的模型。
示例
请求
curl http://localhost:11434/api/copy -d '{
"source": "llama3.2",
"destination": "llama3-backup"
}'
响应
如果成功,则返回 200 OK,如果源模型不存在,则返回 404 Not Found。
删除模型
DELETE /api/delete
删除模型及其数据。
参数
model
:要删除的模型名称
示例
请求
curl -X DELETE http://localhost:11434/api/delete -d '{
"model": "llama3:13b"
}'
响应
如果成功,则返回 200 OK,如果要删除的模型不存在,则返回 404 Not Found。
拉取模型
POST /api/pull
从 ollama 库下载模型。取消的拉取将从上次中断的位置继续,并且多个调用将共享相同的下载进度。
参数
model
:要拉取的模型的名称insecure
:(可选)允许与库建立不安全的连接。仅在开发过程中从自己的库中提取时才使用此项。stream
:(可选)如果false
响应将作为单个响应对象返回,而不是对象流
示例
请求
curl http://localhost:11434/api/pull -d '{
"model": "llama3.2"
}'
响应
如果stream
未指定,或设置为true
,则返回 JSON 对象流:
第一个对象是清单:
{
"status": "pulling manifest"
}
然后是一系列的下载响应。在完成任何下载之前,completed
可能不包括 key。要下载的文件数取决于清单中指定的层数。
{
"status": "downloading digestname",
"digest": "digestname",
"total": 2142590208,
"completed": 241970
}
下载所有文件后,最终响应为:
{
"status": "verifying sha256 digest"
}
{
"status": "writing manifest"
}
{
"status": "removing any unused layers"
}
{
"status": "success"
}
如果stream
设置为 false,则响应是单个 JSON 对象:
{
"status": "success"
}
推送模型
POST /api/push
将模型上传到模型库。需要先注册 ollama.ai 并添加公钥。
参数
model
:要推送的模型名称,形式为<namespace>/<model>:<tag>
insecure
:(可选)允许与库建立不安全的连接。仅在开发期间推送到库时使用此选项。stream
:(可选)如果false
响应将作为单个响应对象返回,而不是对象流
示例
请求
curl http://localhost:11434/api/push -d '{
"model": "mattw/pygmalion:latest"
}'
响应
如果stream
未指定,或设置为true
,则返回 JSON 对象流:
{ "status": "retrieving manifest" }
然后:
{
"status": "starting upload",
"digest": "sha256:bc07c81de745696fdf5afca05e065818a8149fb0c77266fb584d9b2cba3711ab",
"total": 1928429856
}
然后是一系列上传响应:
{
"status": "starting upload",
"digest": "sha256:bc07c81de745696fdf5afca05e065818a8149fb0c77266fb584d9b2cba3711ab",
"total": 1928429856
}
最后,上传完成后:
{"status":"pushing manifest"}
{"status":"success"}
如果stream
设置为false
,则响应是单个 JSON 对象:
{ "status": "success" }
生成嵌入
POST /api/embed
从模型生成嵌入
参数
model
:要从中生成嵌入的模型的名称input
:要为其生成嵌入的文本或文本列表
高级参数:
truncate
:截断每个输入的结尾以适合上下文长度。如果出现 error,则返回 errorfalse
和上下文长度。默认为true
options
:Modelfile 文档中列出的其他模型参数,例如temperature
keep_alive
:控制模型在请求后加载到内存中的时间(默认值:5m
)
示例
请求
curl http://localhost:11434/api/embed -d '{
"model": "all-minilm",
"input": "Why is the sky blue?"
}'
响应
{
"model": "all-minilm",
"embeddings": [[
0.010071029, -0.0017594862, 0.05007221, 0.04692972, 0.054916814,
0.008599704, 0.105441414, -0.025878139, 0.12958129, 0.031952348
]],
"total_duration": 14143917,
"load_duration": 1019500,
"prompt_eval_count": 8
}
请求 (多路输入)
curl http://localhost:11434/api/embed -d '{
"model": "all-minilm",
"input": ["Why is the sky blue?", "Why is the grass green?"]
}'
响应
{
"model": "all-minilm",
"embeddings": [[
0.010071029, -0.0017594862, 0.05007221, 0.04692972, 0.054916814,
0.008599704, 0.105441414, -0.025878139, 0.12958129, 0.031952348
],[
-0.0098027075, 0.06042469, 0.025257962, -0.006364387, 0.07272725,
0.017194884, 0.09032035, -0.051705178, 0.09951512, 0.09072481
]]
}
列出正在运行的模型
GET /api/ps
列出当前加载到内存中的模型。
示例
请求
curl http://localhost:11434/api/ps
响应
将返回单个 JSON 对象。
{
"models": [
{
"name": "mistral:latest",
"model": "mistral:latest",
"size": 5137025024,
"digest": "2ae6f6dd7a3dd734790bbbf58b8909a606e0e7e97e94b7604e0aa7ae4490e6d8",
"details": {
"parent_model": "",
"format": "gguf",
"family": "llama",
"families": [
"llama"
],
"parameter_size": "7.2B",
"quantization_level": "Q4_0"
},
"expires_at": "2024-06-04T14:38:31.83753-07:00",
"size_vram": 5137025024
}
]
}
生成嵌入
注意:此终端节点已被
/api/embed
POST /api/embeddings
从模型生成嵌入
参数
model
:要从中生成嵌入的模型的名称prompt
:要为其生成嵌入的文本
高级参数:
options
:Modelfile 文档中列出的其他模型参数,例如temperature
keep_alive
:控制模型在请求后加载到内存中的时间(默认值:5m
)
示例
请求
curl http://localhost:11434/api/embeddings -d '{
"model": "all-minilm",
"prompt": "Here is an article about llamas..."
}'
响应
{
"embedding": [
0.5670403838157654, 0.009260174818336964, 0.23178744316101074, -0.2916173040866852, -0.8924556970596313,
0.8785552978515625, -0.34576427936553955, 0.5742510557174683, -0.04222835972905159, -0.137906014919281
]
}
版本
GET /api/version
检索 Ollama 版本
示例
请求
curl http://localhost:11434/api/version
响应
{
"version": "0.5.1"
}