Models

Magic

AI lab building long-context coding models and an autonomous software engineer, researching ultra-long context windows for reasoning over entire codebases at once.

Mistral Large

Mistral AI's flagship reasoning and coding model, alongside the open-weight Mixtral mixture-of-experts line that put the French lab on the map before Mistral Large existed.

Qwen2.5-Coder

Alibaba's open-weight code model family, with the 32B variant matching GPT-4o on several coding benchmarks while staying free to self-host under an Apache 2.0 license.

Llama

Meta's open-weight foundation model family, the base that Code Llama and a huge share of the open-source LLM ecosystem has been built on and fine-tuned from.

Grok

xAI's model family, integrated into X and available via API, with coding and reasoning variants competing directly with GPT, Claude, and Gemini at the frontier tier.

Code Llama

Meta's open-source code model built on Llama 2. Archived July 2025, superseded by newer Llama model generations.

GPT (OpenAI)

OpenAI's GPT family of frontier models, available through ChatGPT and the OpenAI API, including coding-tuned Codex variants.

Gemini

Google's multimodal Gemini model family with long context windows, available in the Gemini app, Google AI Studio and Vertex AI.

Grok

xAI's Grok model family and chatbot, available in the Grok app, on X and through the xAI API, with strong reasoning, real-time search and coding models.

Grok Code Fast 1

xAI's fast, low-cost agentic coding model, built for tool use in IDEs and agents, available through the xAI API and partner editors.

Kimi K2

Moonshot AI's open-weight mixture-of-experts model line, strong at agentic coding and tool use, accessible via the Kimi app and API.

GLM

Z.ai's GLM model family, with open-weight models strong at coding and agentic tasks.

MiniMax

MiniMax's language models, with open-weight releases aimed at coding and agent workflows.

Gemma

Google's family of lightweight open-weight models for running on laptops, phones and the cloud.

Devstral

Mistral's open model built for agentic software engineering tasks.

gpt-oss

OpenAI's open-weight reasoning models that you can run locally or self-host.

Cohere Command

Cohere's enterprise LLM family, tuned for RAG, tool use and multilingual agents.

Qwen3-Coder

Alibaba's open-weight code model family built for agentic coding and long-context repository work.

Llama Guard

Meta's safety classifier models for moderating LLM prompts and responses, part of Purple Llama.

LM Evaluation Harness

EleutherAI's standard framework for benchmarking language models across hundreds of tasks.

Voyage AI

Embedding and reranking models tuned for retrieval, code and domain-specific search.

Jina AI

Search foundation models: embeddings, rerankers and a reader API for feeding web content to LLMs.

Salesforce CodeGen

Open-source code-generation model family from Salesforce Research, trained up to 16B parameters, with CodeGen2.5 matching larger models at 7B.

StarCoder

Open-access code model family from the BigCode project, with StarCoder2 in 3B, 7B, and 15B sizes, trained on The Stack v2 across 600+ languages.

Phi-3 Code

Microsoft's small language model family, with Phi-3-mini delivering GPT-3.5-level capability at 3.8B parameters. Superseded by Phi-4 and later releases.

Codestral

Mistral's 22B-parameter code model with a 256k context window and fill-in-the-middle completion. Relicensed to Apache 2.0 in April 2026.

Claude

Anthropic's AI assistant and model family, used for coding, data analysis, writing, and research. Claude Code ships with the Pro plan for agentic coding.

DeepSeek

Chinese AI lab's open-weight model family, including V3 for general tasks, R1 for reasoning, and a dedicated Coder line, free to self-host or use commercially.