Models

14B compact Qwen 3 model for efficient local deployment.

Alibaba's frontier open-weight MoE model with hybrid thinking.

32B Qwen 3 model offering strong reasoning at mid-size cost.

72B dense open-weight model with hybrid thinking from Alibaba.

Cost-efficient Nova 2 model with 1M context.

Next-generation Nova flagship with 1M context from Amazon Bedrock.

Low-cost multimodal model from Amazon for high-throughput workloads.

Ultra-low-cost text-only model from Amazon.

Amazon's most capable multimodal model, available through Amazon Bedrock.

Previous Anthropic flagship, now superseded by Claude Opus 4.

Fast, low-cost model with stronger capabilities than Claude 3 Haiku.

Mid-2024 release setting a new standard for coding and reasoning at mid-tier price.

Anthropic Claude 3.7 Sonnet.

Fast, low-cost Claude model for latency-sensitive workloads.

Anthropic's most capable model for complex reasoning and long-context work.

Most capable Claude Opus model.

Anthropic Claude Opus 4.5.

Anthropic Claude Opus 4.6.

Anthropic Claude Opus 4.7.

Balanced mid-tier Claude model with strong general capability and price.

Balanced performance and speed.

Anthropic Claude Sonnet 4.6.

Cohere's most capable model with 256K context, optimized for enterprise agentic tasks.

Efficient mid-size model from Cohere for RAG and agentic tasks.

Cohere's flagship model optimized for enterprise RAG and complex tasks.

Open-weight reasoning model matching o1 performance, fully open-source.

Open-weight frontier model competitive with GPT-4o and Claude Sonnet at fraction of training cost.

Latest version of DeepSeek V3.

Fast multimodal model from the Gemini 1.5 generation.

Previous Google flagship with 1M context window, superseded by Gemini 2.

Low-latency, low-cost multimodal model with 1M context.

Google's flagship multimodal model with very long context.

Fast and efficient multimodal model.

Ultra-fast lightweight variant.

Google Gemini 2.5 Pro — state-of-the-art thinking model.

Open-weights 27B model from Google with state-of-the-art performance at its size.

Open-weights 9B model from Google, competitive with much larger models.

Open-weights 70B model for high-quality general use.

Smaller open-weights Llama for on-device and cost-sensitive use.

Meta's largest open-weights model, competitive with frontier closed models.

Updated 70B open-weights model with 128k context and improved tool calling.

Multimodal 11B model from Meta supporting text and image inputs.

Small on-device model for edge and mobile deployments.

Meta Llama 3.3 70B — improved instruction-following.

High-performance multimodal model.

Efficient multimodal model with 17B active parameters.

3.8B instruction-following model targeting mobile and edge deployment.

14B small language model from Microsoft Research with state-of-the-art STEM reasoning.

Compact yet capable small language model.

14B reasoning-specialized Phi model with extended thinking.

15B multimodal reasoning model with image understanding.

Mistral's code-specialized model with long context.

Specialized code generation model.

Agentic coding model for software development.

Compact open-weights model that outperforms Llama 2 13B on many benchmarks.

Mistral's flagship commercial model with tool calling and structured outputs.

Top-tier reasoning and coding model.

Balanced performance and cost.

12B open-weights model built with NVIDIA, with 128k context.

Efficient open-weights mid-sized model from Mistral.

Fast and affordable.

Open-weights mixture-of-experts model with GPT-3.5 class performance.

NVIDIA-tuned Llama 3.1 70B with state-of-the-art alignment and helpfulness.

49B parameter efficient model with frontier reasoning capability from NVIDIA.

Previous-gen GPT-4 flagship with 128k context, now superseded by GPT-4o.

OpenAI GPT-4.1

Smaller, faster and cheaper version of GPT-4.1.

Ultra-fast nano variant of GPT-4.1.

Fast, multimodal model for general use with 128k context.

Low-cost, fast multimodal model for high-volume tasks.

OpenAI's frontier flagship model with long context and advanced reasoning.

OpenAI GPT-5.1.

OpenAI GPT-5.2.

OpenAI GPT-5.3.

OpenAI GPT-5.4.

Cost-efficient variant of GPT-5.4.

Ultra-fast nano variant of GPT-5.4.

Reasoning-focused model that thinks before answering.

Smaller o1-series reasoning model, now superseded by o3-mini.

OpenAI's most powerful reasoning model, successor to o1.

Cost-efficient reasoning model with strong STEM performance.

Highest capability reasoning model.

Fast reasoning model.

Previous generation Grok model, superseded by Grok 3.

xAI's frontier reasoning model with real-time web access.

Fast and cost-efficient reasoning model.

Most capable Grok model.

High-speed variant of Grok 4.