DeepSeek
generalOpen-weight LLMs including a 671B MoE model that matches GPT-4o at far lower API cost
This category covers tools built around large language models: infrastructure for deploying, fine-tuning, evaluating, and monitoring LLMs in production. With 369 tools listed, it is one of the more technical categories on the site, aimed primarily at developers and ML engineers rather than end users.
Open-weight LLMs including a 671B MoE model that matches GPT-4o at far lower API cost
Create slides from text instantly
Long-form writing editor for storytellers
Convert video to 3D animation instantly
Extract text from images with OCR
Create and schedule AI photos, videos, and voiceovers
Text-to-podcast tool with 120+ AI voices and natural conversations in multiple languages
Rewrite AI text to bypass detection
Humanize AI text to bypass detection
Directory of AI video creation tools with workflows
Convert AI-generated text to human-sounding writing
Convert PDFs into TikTok-style study videos
Extract text from images, documents, and screenshots
Create custom emojis with AI
Voice-to-text dictation and transcription
Extract summaries and key points from text
Create, translate, and resize on-brand images in 130+ languages
Information resource for craftui
Humanize AI writing to sound natural
AI photo studio that creates professional product photos from simple images
Convert handwritten documents to digital text
Text-to-audio conversion with multiple languages and voice options
Text-to-video and image-to-video AI
Convert text prompts into scalable vector designs and illustrations
LLM tooling has exploded alongside the models themselves, and the category now spans several distinct problem areas. Deployment and serving tools like PeriFlow and Dstack help teams run models efficiently at scale. Evaluation and observability tools like UpTrain and AIWatch track model quality, drift, and cost over time. Memory and retrieval tools like Cognee add persistent context or RAG capabilities to LLM applications. When choosing, the key questions are infrastructure fit (cloud, on-prem, or hybrid), model compatibility (OpenAI-only vs. open-weight models), and whether the tool addresses your actual bottleneck, whether that is latency, cost, accuracy, or developer velocity. Pricing structures vary: some tools charge per token processed, others per seat or per API call. Open-source options exist across most sub-categories, which is worth considering for teams with engineering capacity to self-host.