large language

Cogito v2 671B

SOTA 671B MoE model trained for better outputs from shorter reasoning chains

‌

Model details

Developed by
Deep Cogito
Model family
Cogito
Use case
large language
Version
v2
Size
671B
License
MIT
Readme
View

Example usage

Cogito v2 671B, currently in preview, is a frontier LLM that offers SOTA intelligence with shorter reasoning chains thanks to Iterated Distillation & Amplification (IDA), a novel research technique that distills improvements from inference-time reasoning back into the model weights.

Thanks to IDA, Cogito models arrive at strong results using fewer reasoning tokens, improving their cost and speed in real-world agents and applications.

Input

1from openai import OpenAI
2import os
3
4model_url = "" # Copy in from API pane in Baseten model dashboard
5
6client = OpenAI(
7    api_key=os.environ['BASETEN_API_KEY'],
8    base_url=model_url
9)
10
11# Chat completion
12response_chat = client.chat.completions.create(
13    model="",
14    messages=[
15        {"role": "user", "content": "Write FizzBuzz."}
16    ],
17    temperature=0.6,
18    max_tokens=1000,
19)
20print(response_chat)

JSON output

1{
2    "id": "143",
3    "choices": [
4        {
5            "finish_reason": "stop",
6            "index": 0,
7            "logprobs": null,
8            "message": {
9                "content": "[Model output here]",
10                "role": "assistant",
11                "audio": null,
12                "function_call": null,
13                "tool_calls": null
14            }
15        }
16    ],
17    "created": 1741224586,
18    "model": "",
19    "object": "chat.completion",
20    "service_tier": null,
21    "system_fingerprint": null,
22    "usage": {
23        "completion_tokens": 145,
24        "prompt_tokens": 38,
25        "total_tokens": 183,
26        "completion_tokens_details": null,
27        "prompt_tokens_details": null
28    }
29}