AI model library

DeepSeek-V4.1-Flash-1017 is a closed-weights large language model from DeepSeek, released on 2026-10-17 as a Flash variant of the V4.1 family.

DeepSeek's large language model supporting text and image input, excels at reasoning and agentic coding with a 1M token context window.

A fast variant in the M3.1 series released by MiniMax on October 19, 2026. Further details on capabilities are not available from the collected facts.

NVIDIA's open-weight vision-language reasoning model handling text, image, and video inputs with a 1M-token context window, designed for multimodal understanding and agentic tool use.

Zhipu AI's flagship LLM for long-horizon coding, agents, and complex project delivery with a 1M token context window.

An embedding model from Google in the Gemma family, released October 2026 with closed weights.

Moonshot AI's Kimi K3, a multimodal model with 1M context window and toggleable max-effort thinking for long-horizon agent work.

StepFun's Step 5 Preview is a closed-source large language model focused on delivering high-quality conversation experiences.

Alibaba Qwen's 27B parameter large language model, closed-source.

Open-weight vision-language model from Alibaba's Qwen team, released April 2025. Accepts text and image input with a 131,072-token context window, supporting chat, reasoning, tool use, and self-hosted deployment.

Qwen3.5-9B is a 9B-scale large language model from Alibaba's Qwen team, released on 2026-10-09 with closed weights, intended for general language tasks such as chat.

Zhipu AI's flagship GLM model with million-token context window and tool use capabilities for long-horizon coding agents.

MiniMax's H3 is a closed-source large language model set to be released on September 29, 2026.

Qwen3 14B is an open-weights instruction model from Alibaba's Qwen team for multilingual chat, reasoning, and tool use, with a 131,072-token context window and text-in/text-out.