Models

Model Milestones

• InternLM3: Our latest series of large language models with exceptional reasoning capabilities, leading the open-source models in this class.
• InternVL2.5: Our multimodal large model series, the first open-source model to match the performance of top proprietary models like GPT-4 and Gemini Pro.
• InternLM2.5: The previous generation of models we continue to maintain, offering highly stable and excellent performance after multiple iterations. Available in 7B and 20B model sizes, supporting 1M context length and enhanced instruction-following and tool invocation abilities.

Using the Latest Model

For multi-turn conversation APIs, use the following in the model field:

• internlm3-latest will automatically point to the latest version in the large language model series.
• internvl2.5-latest will automatically point to the latest version in the multimodal model series.

You can check the specific model version they point to by checking the ref_model field in the response from the models API (e.g., internlm3-8b-instruct).

Model List

Large Language Models

Model	Description	Window Length
internlm3-latest	Default points to the latest released InternLM3 model series, currently pointing to internlm3-8b-instruct	32K
internlm3-8b-instruct	Our latest model with leading reasoning capabilities in its scale	32K
internlm2.5-latest	Default points to the latest released InternLM2.5 model series, currently pointing to internlm2.5-20b-chat	32K

Multimodal Models

Model	Description	Context Length
internvl-latest	Points to our latest release in the InternVL series, currently internvl2.5-78b	32K
internvl2.5-latest	Points to our latest release in the InternVL2.5 series, currently internvl2.5-78b	32K
internvl2.5-78b	Our latest multimodal large model, matching the performance of top proprietary models	32K

Request Examples

(1) Python Example

import requests
import json

url = 'https://chat.intern-ai.org.cn/api/v1/models'
header = {
    'Content-Type': 'application/json',
    "Authorization": "Bearer eyJ0eXBlIjoiSl...Please provide the correct token!"
}
data = {}

res = requests.get(url, headers=header, data=json.dumps(data))
print(res.status_code)
print(res.json())
print(res.json()["data"])

(2) OpenAI SDK Example

from openai import OpenAI

client = OpenAI(
    api_key="eyJ0eXBlIjoiSl...Please provide the correct token!",  # Provide token here, without 'Bearer'
    base_url="https://chat.intern-ai.org.cn/api/v1/",
)

models = client.models.list()
for model in models:
    print(model)

(3) OpenAI CLI Example

openai -b "https://chat.intern-ai.org.cn/api/v1/" 
-k "eyJ0eXBlIjoiSl...Please provide the correct token!" 
api models.list

Parameter Explanation

Return Parameters

Parameter	Type	Example	Description
object	string	list	Data type returned, in this case, "list"
data	object	{ "id": "internlm3-latest", "object": "model", "created": 1686935002, "owned_by": "pjlab"}	Model information

Relevant Structure Explanation

• Data Structure Explanation

Parameter	Type	Example	Description
id	string	internlm3-latest	Model ID
object	string	model	Model category, always "model"
created	int	1677652288	Model creation timestamp
owned_by	string	SH-AILab	Model owner, default is "SH-AILab"

// Request
Schema: HTTP
Path:  /api/v1/models
Method: GET
Header: 
    sso-ak: xxxxxx
    Authorization: $BearerToken

// Response
Status Code: 200
Body:
{
    "object": "list",
    "data": [
        {
            "id": "internlm3-latest",
            "object": "model",
            "ref_model": "internlm3-8b-instruct",
            "created": 1722424576,
            "owned_by": "SH-AILab"
        },
        {
            "id": "internvl2.5-latest",
            "object": "model",
            "ref_model": "internvl2.5-78b",
            "created": 1722424576,
            "owned_by": "SH-AILab"
        },
        {
            "id": "internlm2.5-latest",
            "object": "model",
            "ref_model": "internlm2.5-20b-0719",
            "created": 1722424576,
            "owned_by": "SH-AILab"
        }
    ]
}