---

1. Chatterbox — voice cloning & TTS

Replaces: ElevenLabs, PlayHT

Chatterbox is an open-source text-to-speech and voice-cloning model from Resemble AI. It runs locally on your own GPU, clones voices from short samples, and produces natural-sounding speech in multiple languages — no monthly subscription, no voice data on someone else's servers.

Use cases: Podcast intros, voiceovers for shorts, synthetic narrators for explainer videos, dubbing, game audio.

Link: resemble.ai/learn/models/chatterbox

2. DeepSeek — LLM & reasoning

Replaces: ChatGPT (GPT-4/5), Claude

DeepSeek is a family of powerful open-weight models (DeepSeek-V3, DeepSeek-R1) with reasoning capabilities at GPT-4 level. Available both as a free web chat and for self-hosting. Especially strong on code, math, and multi-step reasoning — at a fraction of the API cost.

Use cases: Coding assistant, research, content generation, complex analysis, self-hosted AI workflows without vendor lock-in.

Link: chat.deepseek.com

3. ComfyUI — image generation & workflows

Replaces: Midjourney, Nano Banana, DALL·E, Adobe Firefly

ComfyUI is a node-based interface for Stable Diffusion, Flux, SDXL, and other open-source image models. Instead of a text-prompt field you build visual pipelines: image-to-image, ControlNet, inpainting, LoRAs, upscaling — all in a single graph. Full control over every step, locally on your own hardware, with no credit limits.

Use cases: Product visualisation, concept art, character sheets, marketing assets, iterative image development, video frames for AnimateDiff.

Link: comfy.org

4. AnythingLLM — knowledge assistant for documents

Replaces: NotebookLM, ChatGPT Projects, Mem.ai

AnythingLLM is an open-source desktop and Docker app that lets you build your own RAG-based knowledge bases from PDFs, websites, Notion exports, and other sources. Fully local with Ollama, or connect to OpenAI/Anthropic. Multi-user, workspaces, agent support — and your data stays with you.

Use cases: Personal second brains, team knowledge bases, customer-support bots from documentation, research assistants for confidential content.

Link: anythingllm.com

5. Whisper — speech-to-text & transcription

Replaces: Otter.ai, Rev, Descript Transcription

Whisper is OpenAI's open-source speech-to-text model and the de-facto standard for transcription. Runs locally (e.g. via whisper.cpp or faster-whisper), transcribes in 99+ languages, auto-translates to English, and delivers word-level timestamps — free, offline, no minute caps.

Use cases: Meeting transcripts, podcast subtitles, YouTube caption files, interview analysis, making voice memos searchable.

Link: whisperai.com

Summary

Tool	Replaces	Category
Chatterbox	ElevenLabs	Voice cloning / TTS
DeepSeek	ChatGPT	LLM / reasoning
ComfyUI	Midjourney, Nano Banana	Image generation
AnythingLLM	NotebookLM	Knowledge RAG
Whisper	Otter	Transcription

All five are production-ready today. If you don't mind a bit of setup, you'll save easily a four-figure sum per year — and keep full control over data, workflows, and output.