196 results
Foundation models you can call or deploy
by Moonshot AI·Jul 2026
2.8T-parameter natively-multimodal MoE with 1M-token context from Moonshot AI; built for long-horizon coding, knowledge work, and deep reasoning.
by Inflection AI·Mar 2024
Inflection AI's foundation model powering the Pi assistant, achieving 94% of GPT-4 performance at 40% of the training compute with major coding/math improvements.
by Writer·Oct 2024
Enterprise-grade frontier LLM trained primarily on synthetic data, with strong tool calling, built-in RAG, code generation, and 128K context window.
by Amazon·Dec 2024
Amazon's cost-effective multimodal model optimized for speed, processing text, images, and video inputs at low latency.
by Suno·Nov 2024
Suno's AI music generation model producing full-length songs with vocals, instrumentals, and lyrics from text prompts.
by Technology Innovation Institute·May 2024
Technology Innovation Institute's open-source multilingual model trained on 5.5 trillion tokens with strong performance efficiency.
by Meta AI·Jun 2023
Meta's open-source music generation model that creates high-quality music from text descriptions or melody conditioning.
by Runway·Jun 2024
Runway's video generation model producing high-fidelity 10-second video clips from text and image prompts with temporal consistency.
by Amazon Web Services·Apr 2025
Amazon's most capable multimodal model for complex reasoning tasks, multi-step agentic collaboration, and model distillation with 1M token context window.
by MiniMax·Jan 2025
456B parameter MoE model (45.9B active) with up to 4M token inference context using hybrid Lightning+Softmax attention. Open-weight.
by MiniMax·Jun 2026
MiniMax's flagship LLM featuring proprietary MSA (MiniMax Sparse Attention) architecture with 1M token context and native multimodality, open-sourced.
by Databricks·Mar 2024
Databricks' open-source mixture-of-experts LLM with 132B total parameters, excelling at code and general language tasks.
by RWKV Foundation·Jun 2024
Linear attention architecture model combining RNN efficiency with Transformer quality, enabling infinite context length at constant memory.
by Shanghai AI Lab·Jul 2024
Shanghai AI Lab's open-source model excelling at reasoning, long-context understanding, and tool use with 1M token context support.
by Technology Innovation Institute·Dec 2024
TII's latest open-source LLM with improved reasoning and multilingual capabilities, trained on curated high-quality data.
by Google DeepMind·Aug 2024
Google DeepMind's highest-quality text-to-image model with photorealistic output, rich detail, and fewer artifacts.
by Snowflake·Apr 2024
480B total parameter Dense-MoE Hybrid model (17B active) excelling at SQL generation, coding, and instruction following. Trained for under $2M, Apache 2.0 licensed.
by Reka AI·Apr 2024
Frontier-class multimodal language model supporting text, images, video, and audio inputs as Reka AI's top-tier offering.
Amazon's multimodal foundation model for complex tasks including text, image, and video understanding via Amazon Bedrock.
by Kuaishou·Oct 2024
Kuaishou's video generation model producing high-quality videos up to 2 minutes with improved motion and physics simulation.
by Midjourney·Dec 2023
Major model update with improved prompt understanding, coherence, text rendering in images, and photorealistic capabilities.
by OpenAI·Apr 2022
OpenAI's image generation model using diffusion with CLIP guidance, supporting image creation, editing, and variations.
by Adobe·Apr 2024
Adobe's commercially-safe generative AI model for high-quality image creation with improved photorealistic output and style control.
by Cohere·Mar 2025
Cohere's multimodal embedding model supporting text and images for semantic search and retrieval applications.
