Supported AI Models

Geekflare Connect is a Bring Your Own Key (BYOK) platform providing access to over 35 LLMs from the world’s leading providers. Unlike the all-in-one Geekflare Chat subscription, Connect allows you to use your own API keys for direct billing and control. This list is continually updated as new models are released. If you don’t see a model you need, please let us know!

OpenAI

OpenAI is renowned for its powerful and versatile GPT (Generative Pre-trained Transformer) models, setting the industry standard for complex reasoning, instruction following, and natural language generation.

Model Name	Best for
GPT 5.4 Pro	Higher reasoning and accuracy for complex, demanding tasks.
GPT 5.4	General purpose, versatile model that can handle software development, writing, reasoning and conversations.
GPT 5.3 Instant	Optimized for everyday conversations, more contextualized web search results and directly helpful answers with fewer disclaimers.
GPT 5.2 Codex	An optimized model for long-horizon agentic coding, featuring context compaction and improved Windows support. It excels at complex refactors and high-efficiency tasks.
GPT 5.2	Reasoning and high speed. Useful for coding and agentic tasks.
GPT 5.2 Chat	General conversations and every day tasks. The same as the one used in ChatGPT 5.2.
GPT 5.2 Pro	High intelligence, deep reasoning and precise responses.
GPT 5.1 Instant	Fast, enhanced GPT-5 model for everyday tasks and quick interactions.
GPT 5.1 Thinking	Flagship OpenAI GPT-5 model optimized for deep reasoning and adaptive task performance.
GPT 5.1 Codex	Advanced GPT-5.1 coding model optimized for high-accuracy software generation, debugging, and complex code reasoning.
GPT 5.1 Codex Mini	Lightweight, fast GPT-5.1 coding model designed for quick code tasks with low latency.
GPT 5	Predecessor of GPT 5.1, suited for coding and complex reasoning.
GPT 5 Pro	Premium GPT-5 model offering enhanced reasoning, depth, and task reliability for professional-grade tasks.
GPT 5 Chat	Chat-optimized version of GPT 5 model used within ChatGPT. Used for general-purpose tasks like brainstorming, problem-solving etc.
GPT 5 mini	Balanced model for everyday tasks. This is cheaper than GPT-5 but retains strong reasoning, like drafting blog posts, summarizing, creating simple code snippets, etc.
GPT 5 nano	Lightweight, cheap GPT 5 model for simple and high-volume tasks like data cleaning, bulk text generation.
GPT 4.1	This non-reasoning model is suited for general tasks that don’t require much analysis, like drafting professional emails and social media posts.
GPT 4.1 mini	This is a lighter version of 4.1 that gives quick responses. Suited for chatbots and customer support replies.
GPT 4.1 nano	This is an even lighter and quicker version of 4.1 with a lesser knowledge base. It’s suited for simple tasks like data cleaning, correcting punctuation and grammar, and simple Q&A, like definitions.
GPT 4o	Flexible multimodal model for real-time text, vision, and audio tasks.
GPT 4o mini	Fast multimodal model for simple image Q&A, basic chat.
o4 mini	Affordable reasoning model for coding and visual tasks. Predecessor of GPT 5 mini.
o4 mini Deep Research	Affordable deep research model for multi-step research, ability access data from the internet as well as your own data through MCP connectors.
o3	Predecessor to GPT 5 for complex reasoning tasks in math, science, coding, and visual.
o3 Deep Research	Analyze hundreds of sources into a rich report with citations, creating long-form content from extensive online research.
o3 mini	Cost-efficient model for STEM and coding-based tasks.

Google Gemini

Google’s Gemini family of models is built from the ground up to be multimodal, excelling at processing and reasoning across text, images, code, and video.

Model Name	Best for
Gemini 3.1 Flash Lite	Fast, cost-efficient model optimized for high volume tasks like classification, translation etc.
Gemini 3.1 Pro	Improved token efficiency, optimized for software engineering and agentic tasks.
Gemini 3 Flash	Improved reasoning and intelligence at lower and cost and higher speed.
Gemini 3 Pro	Flagship model of Gemini with configurable thinking levels to balance complex reasoning and speed.
Gemini 3 Pro Image	Google Nano Banana Pro generates 4K images with superior speed and contextual understanding.
Gemini 2.5 Pro	High performance with enhanced thinking, multimodal understanding, processing large datasets, mathematical problem solving, and complex coding.
Gemini 2.5 Flash	Balanced model for everyday usage, high-volume tasks with thinking capabilities. Good for virtual assistants, real-time summarization, and customer-facing agents.
Gemini 2.5 Flash Image	Nano Banana is Google code name for Gemini 2.5 Flash Image, a specialized model for natural language photo generation.
Gemini 2.5 Flash Lite	Cost efficient, lightweight model for quick response. Good for summarizing, and processing large datasets.
Gemini 2.0 Flash	Older model, suited for daily simple tasks not requiring advanced thinking, like basic chat and light content generation.

Anthropic

Anthropic Claude models offers top-tier performance with a strong reasoning and coding.

Model Name	Best for
Claude Sonnet 4.6	Balanced model for cost, speed and intelligence with a 1M context window.
Claude Opus 4.6	High intelligence for coding and building agents with a 1M context window.
Claude Opus 4.5	Premium coding model to give maximum intelligence with improved performance.
Claude Opus 4.1	Superior performance, complex reasoning, advanced coding, and agentic tasks.
Claude Opus 4	Sustained performance on long-running tasks, coding, and agent workflows.
Claude Sonnet 4.5	Latest model with highest intelligence, exceptional agent and coding capabilities
Claude Sonnet 4	Hybrid reasoning model for high-volume uses like customer-facing chat agents, and visual data extraction
Claude Haiku 4.5	High performance & fast at one-third the cost and twice as fast as Sonnet 4.
Claude 3.5 Haiku	Fast performance, code completion, customer service chatbots, data extraction, real-time content moderation

Mistral AI

Mistral AI is an efficient language model that offer competitive performance with a focus on developer-friendliness and customization.

Model	Best for
Mistral Medium	High performance at lower cost for enterprise knowledge base and workflows
Mistral Codestral	AI-powered software development for enterprises

Perplexity

Perplexity models are designed as “answer engines,” excelling at providing accurate, real-time, and cited information by integrating web search capabilities directly into the language model.

Model	Best for
Sonar	Lightweight model for quick responses, general queries.
Sonar Pro	Advanced search, complex queries, follow-ups.
Sonar Reasoning	Fast reasoning model for problem-solving, powered by search.
Sonar Deep Research	Conducting exhaustive web search, processing multiple information sources and generating detailed reports with citations.

xAI Grok

Grok is developed by xAI with the goal of creating AI that can understand the universe. It is known for its real-time knowledge of the world via the 𝕏 (formerly Twitter) platform and its unique, rebellious personality.

Model	Best for
Grok 4.1 Fast Reasoning	Lower cost reasoning model. Ability to give better emotional and creative responses without compromising on speed.
Grok 4.1 Fast non-reasoning	Low cost model for every day tasks like summarizing, high-volume tasks and chatbots.
Grok 4	Flagship model of xAI—high performance in math, reasoning, and natural language. Good for technical research, coding, document/image/voice processing
Grok 4 Fast	Cost-efficient quick response model with limited reasoning abilities. Suited for high-volume chatbots, summarizing. Succeeded by Grok 4.1 Fast non-reasoning.
Grok 4 Fast Reasoning	Cost-efficient reasoning model. Suitable for mathematical deductions, multistep logical inferences. Succeeded by Grok 4.1 Fast reasoning
Grok 3	Non-reasoning model suitable for data extraction, text summarization and coding. Possesses extensive domain knowledge in healthcare, finance, law etc.
Grok 3 Mini	Lightweight reasoning model for tasks where deep domain knowledge is not required. Best for document summarization, mathematical reasoning and logic based tasks

DeepSeek

DeepSeek AI focuses on developing strong models, particularly in the domains of code and mathematics. Their models are an good choice for technical and programming-related tasks.

Model	Best for
DeepSeek Chat	General-Purpose, non-thinking model for conversations, everyday tasks, coding, summarization, and creative writing.
DeepSeek Reasoner	Reasoning model for complex, step-by-step logical reasoning for math problems, scientific reasoning, and code debugging

Don’t See a Model You Need?

If your team relies on a model that isn’t listed here, please contact our support team to submit a request. We are always evaluating new models to add to the Geekflare AI suite.

Getting started

General

Features

Geekflare Connect

Geekflare Chat

Supported AI Models

OpenAI

Google Gemini

Anthropic

Mistral AI

Perplexity

xAI Grok

DeepSeek

Don’t See a Model You Need?

​OpenAI

​Google Gemini

​Anthropic

​Mistral AI

​Perplexity

​xAI Grok

​DeepSeek

​Don’t See a Model You Need?

OpenAI

Google Gemini

Anthropic

Mistral AI

Perplexity

xAI Grok

DeepSeek

Don’t See a Model You Need?