AI model directory

Latest GPT: fast, efficient, strong at visual reasoning and writing

Frontier GPT-6 for complex professional, coding and research work

Fastest and cheapest GPT-6 for everyday and high-volume chat

Frontier model for complex professional, coding and research work

Balances intelligence and cost for everyday production work

Fastest and cheapest GPT-5.6 for high-volume workloads

Previous-generation GPT-5 with 400K context and advanced reasoning

Refined GPT-5 with improved tone and efficiency

Major breakthrough in coding, math, and multi-step reasoning

Most capable Claude: hardest reasoning and long-horizon work (slower, premium)

Current Opus: complex coding, agents and enterprise work at a lower price

Current Sonnet: best speed-to-intelligence balance for everyday work

Fastest, cheapest Claude with 1M context

Most capable Claude: hardest reasoning and long-horizon agentic work

Complex agentic coding and enterprise work with 1M context

Best speed-to-intelligence balance; near-Opus coding quality

Highly autonomous long-horizon agentic work and knowledge tasks

Autonomous agentic work, vision and memory tasks

Frontier coding and agents with 1M context

Fast, cost-efficient Sonnet with adaptive thinking and 1M context

Premium model combining maximum intelligence with practical performance

Fastest and most cost-effective Claude for simple tasks

Strong model for real-world agents and coding

Most capable Gemini: deep reasoning, long context and agentic coding

Latest Flash: fast, low-cost Gemini for everyday and agentic tasks

Gemini balancing speed and intelligence for agentic tasks

Advanced intelligence and agentic coding (preview)

Fast, capable workhorse model with reasoning

Fastest, most cost-effective 3.5 model for high throughput

Low-latency, cost-effective option for high-volume automation

Earlier Gemini 3 Flash preview

Previous-generation lightweight model — GA until 2026-10-16

Latest Grok: agentic tool calling with configurable reasoning

Previous Grok flagship: agentic tool calling with configurable reasoning

Most capable DeepSeek with 1M context and thinking mode

Fast, very low-cost DeepSeek with 1M context and thinking mode

Flagship GLM for long-horizon agentic tasks with 1M context

High-capability GLM with thinking mode and 200K context

Balanced GLM-5 generation model with thinking mode

Faster GLM-5 variant tuned for throughput

Cost-effective GLM with thinking mode and 200K context

Free, fast GLM tier for lightweight tasks

Strong coding and agentic model with 200K context

Previous-generation GLM with 128K context

Lightweight, very low-cost GLM-4.5 variant

Flagship Kimi: native vision, 1M context, always reasoning

Dedicated coding model with reliable long-context instruction following

High-throughput K2.7 Code, ~180 tokens/s output

Vision + text with thinking and non-thinking modes

Flagship Qwen for complex reasoning and agentic work

Balanced Qwen with long context and thinking mode

Fast, low-cost Qwen for high-volume tasks

Omni-modal Qwen handling text, image and audio input