AI Models

Claude Haiku 4.5 matches Sonnet 4's performance on coding, computer use, and agent tasks at substantially lower cost and faster speeds. It delivers near-frontier performance and Claude's unique character at a price point that works for scaled sub-agent deployments, free tier products, and intelligence-sensitive applications with budget constraints.

Claude Opus 4.1 is a drop-in replacement for Opus 4 that delivers superior performance and precision for real-world coding and agentic tasks. Opus 4.1 advances state-of-the-art coding performance to 74.5% on SWE-bench Verified, and handles complex, multi-step problems with more rigor and attention to detail.

Claude Sonnet 4.5 is the newest model in the Sonnet series, offering improvements and updates over Sonnet 4.

DeepSeek-V3.2-Exp is an experimental model introducing the groundbreaking DeepSeek Sparse Attention (DSA) mechanism for enhanced long-context processing efficiency. Built on V3.1-Terminus, DSA achieves fine-grained sparse attention while maintaining identical output quality.

DeepSeek-V3.2-Exp is an experimental model introducing the groundbreaking DeepSeek Sparse Attention (DSA) mechanism for enhanced long-context processing efficiency. Built on V3.1-Terminus, DSA achieves fine-grained sparse attention while maintaining identical output quality.

Gemini 2.5 Flash is a thinking model that offers great, well-rounded capabilities. It is designed to offer a balance between price and performance with multimodal support and a 1M token context window.

Gemini 2.5 Flash Image is the first fully hybrid reasoning model, letting developers turn thinking on or off and set thinking budgets to balance quality, cost, and latency. Upgraded for rapid creative workflows, it can generate interleaved text and images and supports conversational, multi-turn image editing in natural language. It's also locale-aware, enabling culturally and linguistically appropriate image generation for audiences worldwide.

Gemini 2.5 Flash-Lite is a balanced, low-latency model with configurable thinking budgets and tool connectivity (e.g., Google Search grounding and code execution). It supports multimodal input and offers a 1M-token context window.

Gemini 2.5 Pro is the most advanced reasoning Gemini model, capable of solving complex problems. It features a 2M token context window and supports multimodal inputs including text, images, audio, video, and PDF documents.

GLM-4.5 and GLM-4.5-Air are the latest flagship models, purpose-built as foundational models for agent-oriented applications. Both leverage a Mixture-of-Experts (MoE) architecture. GLM-4.5 has a total parameter count of 355B with 32B active parameters per forward pass, while GLM-4.5-Air adopts a more streamlined design with 106B total parameters and 12B active parameters.

As the latest iteration in the GLM series, GLM-4.6 achieves comprehensive enhancements across multiple domains, including real-world coding, long-context processing, reasoning, searching, writing, and agentic applications.

GPT-5 is OpenAI's flagship language model that excels at complex reasoning, broad real-world knowledge, code-intensive, and multi-step agentic tasks.

GPT-5 mini is a cost optimized model that excels at reasoning/chat tasks. It offers an optimal balance between speed, cost, and capability.

GPT-5 nano is a high throughput model that excels at simple instruction or classification tasks.

GPT-5 pro uses more compute to think harder and provide consistently better answers. Since GPT-5 pro is designed to tackle tough problems, some requests may take several minutes to finish.

xAI's latest and greatest flagship model, offering unparalleled performance in natural language, math and reasoning - the perfect jack of all trades.

Grok 4 Fast is xAI's latest multimodal model with SOTA cost-efficiency and a 2M token context window. It comes in two flavors: non-reasoning and reasoning.

Grok 4 Fast is xAI's latest multimodal model with SOTA cost-efficiency and a 2M token context window. It comes in two flavors: non-reasoning and reasoning.

Kimi K2 Thinking Turbo can execute up to 200-300 sequential tool calls without human interference, reasoning coherently across hundreds of steps to solve complex problems. Built as a thinking agent, it reasons step by step while using tools, achieving state-of-the-art performance on Humanity's Last Exam (HLE), BrowseComp, and other benchmarks, with major gains in reasoning, agentic search, coding, writing, and general capabilities. Extremely fast variant of Kimi K2 Thinking.

Kimi K2 Turbo is a large-scale Mixture-of-Experts language model featuring 1 trillion total parameters with 32 billion active per forward pass. It is optimized for agentic capabilities, including advanced tool use, reasoning, and code synthesis. Excels across a broad range of benchmarks, particularly in coding (LiveCodeBench, SWE-bench), reasoning (ZebraLogic, GPQA), and tool-use (Tau2, AceBench) tasks. Extremely fast variant of Kimi K2.

Where performance meets efficiency. This model supports high-performance conversational AI designed for content creation, enterprise applications, and research, offering advanced language understanding capabilities, including text summarization, classification, sentiment analysis, and code generation.

The Llama 4 collection of models are natively multimodal AI models that enable text and multimodal experiences. These models leverage a mixture-of-experts architecture to offer industry-leading performance in text and image understanding.

The Llama 4 collection of models are natively multimodal AI models that enable text and multimodal experiences. These models leverage a mixture-of-experts architecture to offer industry-leading performance in text and image understanding.

Complex thinking, backed by deep understanding, with transparent reasoning you can follow and verify. The model excels in maintaining high-fidelity reasoning across numerous languages, even when switching between languages mid-task.

MiniMax M2 redefines efficiency for agents. It is a compact, fast, and cost-effective MoE model (230 billion total parameters with 10 billion active parameters) built for elite performance in coding and agentic tasks, all while maintaining powerful general intelligence.

Mistral Large is ideal for complex tasks that require large reasoning capabilities or are highly specialized - like Synthetic Text Generation, Code Generation, RAG, or Agents.

Mistral Medium 3 delivers frontier performance while being an order of magnitude less expensive. For instance, the model performs at or above 90% of Claude Sonnet 3.7 on benchmarks across the board at a significantly lower cost.

Mistral Small is the ideal choice for simple tasks that one can do in bulk - like Classification, Customer Support, or Text Generation. It offers excellent performance at an affordable price point.

A 12B model with image understanding capabilities in addition to text.

Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models. Built upon extensive training, Qwen3 delivers groundbreaking advancements in reasoning, instruction-following, agent capabilities, and multilingual support

Efficient coding specialist balancing performance with cost-effectiveness for daily development tasks while maintaining strong tool integration capabilities.

The Qwen3 series Max model has undergone specialized upgrades in agent programming and tool invocation compared to the preview version. The officially released model this time has achieved state-of-the-art (SOTA) performance in its field and is better suited to meet the demands of agents operating in more complex scenarios.

Over the past few months, increasingly clear trends toward scaling both total parameters and context lengths have emerged in the pursuit of more powerful and agentic artificial intelligence (AI). The latest advancements in addressing these demands are centered on improving scaling efficiency through innovative model architecture. This next-generation foundation model is called Qwen3-Next.

Qwen3 series VL models feature significantly enhanced multimodal reasoning capabilities, with a particular focus on optimizing the model for STEM and mathematical reasoning. Visual perception and recognition abilities have been comprehensively improved, and OCR capabilities have undergone a major upgrade.