llmreference
Together AI

Using Dolphin 2.5 Mixtral 8x7B on Together AI

Implementation guide · Dolphin · Cognitive Computations

Serverless

Quick Start

  1. 1
    Create an account at Together AI and generate an API key.
  2. 2
    Use the Together AI SDK or REST API to call dolphin-2.5-mixtral-8x7b — see the documentation for request format.
  3. 3
    You'll be billed $0.60/1M input, $0.60/1M output tokens. See full pricing.

Code Examples

Install
pip install together
API key
TOGETHER_API_KEY
Model ID
dolphin-2.5-mixtral-8x7b

Together uses "organization/model-name" format, e.g. "meta-llama/Llama-4-Scout-17B-16E-Instruct" or "Qwen/QwQ-32B". See the Together model catalog for the exact ID.

from together import Together

client = Together()  # reads TOGETHER_API_KEY from env
response = client.chat.completions.create(
    model="dolphin-2.5-mixtral-8x7b",
    messages=[{"role": "user", "content": "Hello"}]
)
print(response.choices[0].message.content)

About Together AI

Platform for running open-source and proprietary LLMs

Together AI is a platform for running open-source and proprietary LLMs with fast serverless and dedicated endpoints at competitive inference pricing.

Pricing on Together AI

TypePrice (per 1M)
Input tokens$0.60
Output tokens$0.60

Capabilities

No model capability flags are currently sourced.

About Dolphin 2.5 Mixtral 8x7B

The Dolphin 2.5 Mixtral 8x7B is a sophisticated large language model designed primarily for coding tasks, known for its proficiency across diverse programming languages including Kotlin. It utilizes the Mixtral-8x7b architecture and has been fine-tuned on datasets like Dolphin-Coder and MagiCoder, employing qLoRA and Axolotl during training. Featuring a 16k context window for fine-tuning and a base context window of 32k, it offers powerful yet uncensored capabilities, allowing it to handle a wide range of prompts, albeit this introduces ethical considerations. The model is available in various formats on platforms like Hugging Face, catering to different needs with options such as GGUF and GPTQ quantization levels. Despite its strengths, users should be mindful of ethical sensitivities and implement alignment measures when deploying it publicly.

Model Specs

Released2023-12-18
Parameters8x7B
Context32K
ArchitectureMixture of Experts
Knowledge cutoff2023-12

Provider

Together AI
Together AI

San Francisco, California, United States