# Foundation Models > The AI model families themselves: language models like Claude, Gemini, and Gemma, reasoning models like DeepSeek-R1 and o1, plus coding, embedding, speech, and image model weights. One card per family; versions and tiers live inside each card, and the products built on these models live in their own categories. 60 tools in 6 subcategories. Web page: https://ailandscape.org/category/foundation-models · Each tool also has a markdown version at /tool/{slug}.md ## Language Models - [Claude](https://ailandscape.org/tool/claude): Anthropic's family of AI assistants - [Command R+](https://ailandscape.org/tool/command-r): Cohere's RAG-optimized enterprise language model - [DeepSeek-V2](https://ailandscape.org/tool/deepseek-v2): Mixture-of-experts LLM with strong coding and reasoning - [Falcon](https://ailandscape.org/tool/falcon): TII's open-weights language model - [Gemini](https://ailandscape.org/tool/gemini): Google DeepMind's multimodal AI model - [Gemma](https://ailandscape.org/tool/gemma): Google's lightweight open-weights language model - [GPT](https://ailandscape.org/tool/gpt): OpenAI's flagship large language model - [Grok-2](https://ailandscape.org/tool/grok-2): xAI's frontier language model with real-time data access - [Llama](https://ailandscape.org/tool/llama): Meta's open-weights large language model - [Mistral](https://ailandscape.org/tool/mistral): Efficient open-weights language models - [Mistral Large](https://ailandscape.org/tool/mistral-large): Mistral AI's top-tier proprietary language model - [MobileLLM](https://ailandscape.org/tool/mobilellm): Meta's sub-billion parameter models for mobile devices - [Phi-3](https://ailandscape.org/tool/phi-3): Microsoft's small but capable open-weights model - [Pokee Isaac](https://ailandscape.org/tool/pokee-isaac): Agentic 28B model with a 10M-token context window, deployable on one GPU - [Qwen](https://ailandscape.org/tool/qwen): Alibaba's high-performance open-weights language model - [SmolLM](https://ailandscape.org/tool/smollm): Hugging Face's tiny yet capable language model - [TranslateGemma](https://ailandscape.org/tool/translategemma): Google's open translation model family built on Gemma 3, covering 55 languages ## Reasoning & CoT Models - [DeepSeek-R1](https://ailandscape.org/tool/deepseek-r1): Open-weights reasoning model rivaling o1 - [Gemini Thinking](https://ailandscape.org/tool/gemini-thinking): Google's extended-thinking variant of Gemini - [o1](https://ailandscape.org/tool/o1): OpenAI's reasoning model with extended chain-of-thought - [o3-mini](https://ailandscape.org/tool/o3-mini): OpenAI's efficient reasoning model for complex tasks - [QwQ-32B](https://ailandscape.org/tool/qwq-32b): Qwen's open reasoning model with strong math capabilities ## Coding Models - [Alibaba Cloud (Qwen)](https://ailandscape.org/tool/alibaba-cloud-qwen): Alibaba's Qwen family — Qwen3-Coder and Qwen3-Max lead open-weight coding models - [ByteDance](https://ailandscape.org/tool/bytedance): ByteDance's Volcengine AI — Doubao-Seed-Code model integrated into the Trae IDE - [Code World Model](https://ailandscape.org/tool/code-world-model): Meta (Sep 2025) — LLM trained to reason about how code affects program state - [Cognition (SWE-1)](https://ailandscape.org/tool/cognition-swe-1): Creator of SWE-1.5 — a fast-agent frontier model with near-SOTA code performance - [Composer](https://ailandscape.org/tool/composer): Cursor (Oct 2025) — 4x faster than similarly capable models; built for agents - [Devstral 2](https://ailandscape.org/tool/devstral-2): Mistral AI (Dec 2025) — open-source SOTA for code agents; 123B and 24B sizes - [Gemini 3](https://ailandscape.org/tool/gemini-3): Google (Nov 2025) — beats Gemini 2.5 Pro at coding; masters agentic workflows - [GLM](https://ailandscape.org/tool/glm): Z.ai (Jan 2026) — local coding and agentic assistant; the new 30B class standard - [Codex](https://ailandscape.org/tool/codex): OpenAI (Feb 2026) — top agentic coding model; 25% faster, helped create itself - [Grok Code Fast 1](https://ailandscape.org/tool/grok-code-fast-1): xAI (Aug 2025) — economical reasoning model that excels at agentic coding - [JetBrains AI](https://ailandscape.org/tool/jetbrains-ai): JetBrains' Mellum — a 4B parameter open-source model for code completion - [KAT-Dev-72B-Exp](https://ailandscape.org/tool/kat-dev-72b-exp): Kwaipilot (Oct 2025) — open-source 72B SE model; 74.6% on SWE-Bench Verified - [Kimi K2](https://ailandscape.org/tool/kimi-k2): Moonshot AI (Jan 2026) — agentic SOTA; HLE 50.2%, SWE-bench Verified 76.8% - [Llama 4 Maverick](https://ailandscape.org/tool/llama-4-maverick): Meta (Apr 2025) — code-tuned open-weight model - [Mercury Code](https://ailandscape.org/tool/mercury-code): Inception Labs (Nov 2025) — diffusion LLM for coding with 128K context - [Qwen3-Coder](https://ailandscape.org/tool/qwen3-coder): Alibaba Cloud (Jul 2025) — agentic code model; 480B parameters with A35B active - [rnj-1-instruct](https://ailandscape.org/tool/rnj-1-instruct): Essential AI (Dec 2025) — built from scratch for code and STEM; agentic-ready - [Salesforce AI Research](https://ailandscape.org/tool/salesforce-ai-research): Creator of CoDA — a diffusion-based language model for code generation ## Embedding Models - [ALIGN](https://ailandscape.org/tool/align): Google's large-scale visual-language embedding model - [BGE-M3](https://ailandscape.org/tool/bge-m3): Multi-lingual, multi-granularity embedding model by BAAI - [CLIP](https://ailandscape.org/tool/clip): OpenAI's contrastive image-text embedding model - [Cohere Embed](https://ailandscape.org/tool/cohere-embed): Cohere's multilingual text embedding models - [E5-Mistral](https://ailandscape.org/tool/e5-mistral): Instruction-tuned Mistral-based text embeddings - [ImageBind](https://ailandscape.org/tool/imagebind): Meta's unified embedding across 6 modalities - [Jina Embeddings](https://ailandscape.org/tool/jina-embeddings): Long-context text embeddings with 8k token support - [sentence-transformers](https://ailandscape.org/tool/sentence-transformers): Python framework for sentence embeddings - [SigLIP](https://ailandscape.org/tool/siglip): Google's sigmoid loss for image-language pretraining - [text-embedding-ada-002](https://ailandscape.org/tool/text-embedding-ada-002): OpenAI's text embedding model - [Voyage AI](https://ailandscape.org/tool/voyage-ai): State-of-the-art embedding models for retrieval ## Image & Video Models - [DALL-E](https://ailandscape.org/tool/dall-e): OpenAI's image generation model - [Flux](https://ailandscape.org/tool/flux): Black Forest Labs' state-of-the-art image generation model - [Stable Diffusion](https://ailandscape.org/tool/stable-diffusion): Open-weights image generation model ## Speech & Audio Models - [AudioLDM 2](https://ailandscape.org/tool/audioldm-2): Latent diffusion model for audio and music generation - [Bark](https://ailandscape.org/tool/bark): Open-source text-to-audio model by Suno AI - [Kokoro](https://ailandscape.org/tool/kokoro): Lightweight, high-quality open TTS model - [MusicGen](https://ailandscape.org/tool/musicgen): Meta's controllable text-to-music generation model - [StyleTTS2](https://ailandscape.org/tool/styletts2): Human-level TTS via style diffusion and adversarial training - [XTTS](https://ailandscape.org/tool/xtts): Coqui's multilingual voice cloning TTS model --- Part of [AI Landscape](https://ailandscape.org), an open map of the AI ecosystem. Index for AI assistants: https://ailandscape.org/llms.txt