Find AI tools by the job you need done.

Turns geographic map and elevation data into Minecraft Java or Bedrock worlds, with desktop area selection, generation previews, and export.

PrismML’s original Qwen3.6-27B-derived low-bit multimodal family, with 1-bit and ternary weights and local setup instructions under the shared demo’s Bonsai1 guide.

A Cohere Labs 30B-total, 3B-active mixture-of-experts coding model for code generation, agentic software engineering, and terminal tasks, with public BF16 and FP8 Hugging Face weights, Apache 2.0 licensing, and OpenCode or Cohere API try paths listed by official materials.

A Cohere Labs Command A+ model variant with W4A4 quantization, positioned around agentic tool use, multimodal inputs, multilingual work, long context, and Cohere-reported lower hardware requirements.

A newer DeepSeek OCR model release for image/PDF OCR, document-to-Markdown workflows, dynamic resolution, vLLM/Transformers inference, and visual causal flow research.

A DeepSeek model family release positioned around long-context intelligence, reasoning modes, coding benchmarks, and agentic task evaluation.

DeepSeek's 552B-backbone multimodal MoE model with 1M-token context, adjustable reasoning effort, public mixed-precision weights, and reference code for inference, prompt encoding, and evaluation.

A Google DeepMind Gemma 4 model family collection with public checkpoints including Gemma 4 12B, a dense multimodal model Google describes around local agentic workflows, native audio input, and encoder-free vision/audio handling.

A Z.ai flagship text-generation model positioned around 1M-token context, long-horizon coding, agentic engineering, public model weights, API access, and local serving paths.

Z.ai's 753B text-generation model for complex coding and long-horizon agent work, with public weights, an official 1M-token evaluation path, adjustable reasoning effort, and several technical serving options.

A Z.ai 320B-total, 18B-active multimodal mixture-of-experts model for coding, agent workflows, visual input, and long-context work, with a 1M-token context window, public FP8 and BF16 weights, API access, and several technical serving paths.

A 0.9B multimodal OCR model for complex document parsing with several self-hosted or API deployment paths.

A low-bit on-device translation model from AngelSlim, positioned around 33-language offline translation, GGUF access, Android demo use, and 1.25-bit compression.

A Tencent-Hunyuan multilingual translation model family with 1.8B, 7B, and 30B-A3B variants, 33-language support, an AngelSlim 1.25-bit on-device option, IFMTBench, training notes, and several deployment paths.

A Tencent Hy Team 295B-parameter MoE language model for long-context reasoning, coding, tool use, and agent-style workflows, with public Hy3 and Hy3-FP8 weights, Apache-2.0 licensing, and vLLM or SGLang serving notes.

Tencent Hy Team's preview-stage 770B-total, 49B-active Mixture-of-Experts language model for coding, document and analysis work, game development, research, tool use, and long-context tasks, with a 1M-token context window, public BF16 and FP8 weights, and dedicated vLLM or SGLang deployment paths.

A Thinking Machines Lab open-weights multimodal model with text, image, and audio input, adjustable reasoning effort, a 1M-token maximum context window, public BF16 and NVFP4 weights, Tinker fine-tuning access, and deployment recipes for major inference frameworks.

An InternLM model series built to reason across scientific paper pages, molecular and material structures, protein sequences, and physical time-series signals, with current Intern-S2 previews, earlier S1 checkpoints, public weights, and deployment guides.

An NVIDIA Isaac GR00T N1.7 vision-language-action model family for humanoid and generalist robot skills, with a 3B model, post-trained variants, GitHub code, inference and fine-tuning notes, LeRobot-format workflow support, and official robotics developer materials.

A publicly available 8B vision-language interaction model and deployable system that watches a video stream, decides when to speak or stay quiet, and can delegate harder tasks while keeping the stream in view, with time-aligned data, training materials, and vLLM-based setup paths.

Moonshot AI's open-weight native multimodal agentic model with a 2.8T-parameter MoE architecture, 104B activated parameters, a 1M-token context window, public weights, API access, and deployment guidance for long-horizon coding, knowledge work, reasoning, and visual tasks.

A multimodal agentic model positioned around long-horizon coding, tool use, autonomous execution, and broader software workflows.

A Moonshot AI coding-focused agentic model built on Kimi-K2.6, with a 256K context length, image and video input notes, preserve-thinking behavior, multi-step tool-call materials, native INT4 quantization, and deployment paths through vLLM, SGLang, KTransformers, and Moonshot API access.

An NVIDIA kinematic motion diffusion model for generating human and humanoid-robot motion from text plus pose, joint, waypoint, and path constraints, with local inference, a timeline demo, motion exports, official docs, and benchmark materials.