​
Meta Models

Meta develops the Llama series of models, which are high-performance, open-weight large language models optimized for various use cases, including chat and code generation.

Model NameIdentifierDescription
llama-3.3-70b-specdecllama-3.3-70b-specdecSpecialized decoding variant of Llama 3.3 (70B).
llama-3.3-70b-versatilellama-3.3-70b-versatileVersatile version of Llama 3.3 (70B) for various tasks.
llama-3.3-70bllama-3.3-70bStandard Llama 3.3 model with 70B parameters.
llama-3.2-90b-vision-previewllama-3.2-90b-vision-previewVision-enabled variant of Llama 3.2 (90B).
llama-3.2-11b-vision-previewllama-3.2-11b-vision-previewSmaller vision-based variant of Llama 3.2 (11B).
llama-3.2-3b-previewllama-3.2-3b-previewPreview version of Llama 3.2 with 3B parameters.
llama-3.2-1b-previewllama-3.2-1b-previewLightweight preview of Llama 3.2 (1B parameters).
llama-3.1-8b-instantllama-3.1-8b-instantFaster and optimized variant of Llama 3.1 (8B).
llama-3.1-8bllama-3.1-8bStandard Llama 3.1 model with 8B parameters.
llama3-70b-8192llama3-70b-8192Llama 3 variant with 70B parameters and 8192 token context.
llama3-8b-8192llama3-8b-8192Llama 3 variant with 8B parameters and 8192 token context.
llama-guard-3-8bllama-guard-3-8bA specialized Llama 3.3 model focused on content moderation.

​
Mistral Models

Mistral specializes in efficient, state-of-the-art transformer models designed for high-speed inference and strong performance.

Model NameIdentifierDescription
mixtral-8x7b-32768mixtral-8x7b-32768A mixture-of-experts model with 8x7B experts and a 32K token context.

​
DeepSeek Models

DeepSeek focuses on developing cutting-edge language models with optimized performance for reasoning and efficiency.

Model NameIdentifierDescription
deepseek-r1-distill-qwen-32bdeepseek-r1-distill-qwen-32bDistilled version of Qwen-32B optimized by DeepSeek.
deepseek-r1-distill-llama-70b-specdecdeepseek-r1-distill-llama-70b-specdecDistilled version of Llama 70B with specialized decoding.
deepseek-r1-distill-llama-70bdeepseek-r1-distill-llama-70bDistilled Llama 70B model with improved efficiency.

​
Alibaba Qwen Models

Qwen, developed by Alibaba, is designed for code and general-purpose NLP tasks, with a focus on efficiency and accuracy.

Model NameIdentifierDescription
qwen-2.5-coder-32bqwen-2.5-coder-32bA 32B parameter model optimized for code generation.
qwen-2.5-32bqwen-2.5-32bA 32B parameter model optimized for various NLP tasks.

​
Anthropic Models

Anthropic develops the Claude series of AI models, focusing on safety, interpretability, and human-like responses.

Model NameIdentifierDescription
claude-3-7-sonnet-20250219claude-3-7-sonnet-20250219Latest Claude 3.7 Sonnet model, optimized for balanced performance.
claude-3-5-sonnet-20241022claude-3-5-sonnet-20241022Improved version of Claude 3.5 Sonnet.
claude-3-5-haiku-20241022claude-3-5-haiku-20241022Lightweight and fast Claude 3.5 Haiku model.
claude-3-5-sonnet-20240620claude-3-5-sonnet-20240620Earlier version of Claude 3.5 Sonnet.
claude-3-opus-20240229claude-3-opus-20240229Most powerful Claude 3 model for advanced reasoning.
claude-3-sonnet-20240229claude-3-sonnet-20240229Claude 3 Sonnet model with strong performance.
claude-3-haiku-20240307claude-3-haiku-20240307Fastest Claude 3 model for quick responses.

​
OpenAI Models

OpenAI develops the GPT series, including GPT-4o and GPT-4, designed for high-performance natural language understanding and generation.

Model NameIdentifierDescription
gpt-4o-realtime-previewgpt-4o-realtime-previewLatest GPT-4o model with real-time capabilities.
gpt-4o-mini-realtime-previewgpt-4o-mini-realtime-previewA smaller, faster variant of GPT-4o for real-time use.
gpt-4o-latestchatgpt-4o-latestLatest iteration of GPT-4o with enhanced features.
gpt-4o-minigpt-4o-miniA lightweight version of GPT-4o optimized for efficiency.
gpt-4ogpt-4oStandard GPT-4o model with top-tier performance.
gpt-4o-audio-previewgpt-4o-audio-previewGPT-4o variant optimized for audio-related tasks.
gpt-4-turbo-previewgpt-4-turbo-previewA preview of the turbocharged GPT-4 model.
gpt-4-turbogpt-4-turboFaster and cheaper variant of GPT-4.
gpt-4gpt-4Standard GPT-4 model with strong reasoning capabilities.
gpt-3.5-turbogpt-3.5-turboAn efficient version of GPT-3.5 with strong performance.
o3-minio3-miniOptimized small-scale model for fast inference.
o1-minio1-miniMini variant of OpenAI’s “o1” model.
o1o1Standard version of OpenAI’s “o1” model.
o1-previewo1-previewPreview version of OpenAI’s “o1” model.

​
OpenAI Whisper Models

Whisper is OpenAI’s automatic speech recognition (ASR) model, capable of transcribing and translating spoken language.

Model NameIdentifierDescription
distil-whisper-large-v3-endistil-whisper-large-v3-enA distilled, lightweight Whisper model for English transcription.
whisper-large-v3-turbowhisper-large-v3-turboHigh-speed variant of Whisper Large v3.
whisper-large-v3whisper-large-v3Standard Whisper Large v3 model for speech-to-text tasks.
These are the language models support at the moment, please use the identifiers above not the model name.

​
Models Usage

You can use these models in our Query endpoint and OpenAI Compatibility endpoints

​
Query Endpoint Example

curl --request POST \
  --url https://api.wetrocloud.com/v1/query/ \
  --header 'Authorization: Token <api-key>' \
  --header 'Content-Type: application/json' \
  --data '{
    "collection_id": "<collection_id>",
    "request_query": "What are the sales trends for Q1?",
    "model": "claude-3-7-sonnet-20250219"
  }'
OpenAI CompatibilityCreate a Collection