SOHAM FAQ — Frequently Asked Questions

Name: SOHAM
Brand: SOHAM
Availability: InStock
Rating: 4.9 (500 reviews)
Author: Heoster

Question 1

What is SOHAM?

Accepted Answer

SOHAM (Self Organising Hyper Adaptive Machine) is a free, multi-model AI assistant built by Heoster at CODEEX-AI. It gives you access to 35+ AI models from Groq, Google, Cerebras, HuggingFace, and OpenRouter through a single interface — with smart auto-routing, voice features, PDF analysis, image generation, and real-time web search. The name is also inspired by the Sanskrit mantra Soham (So Hum), meaning "I am That."

Question 2

Is SOHAM completely free?

Accepted Answer

Yes — completely free, forever. All core features including 35+ AI models, image generation, web search, voice, and PDF analysis are free with no hidden fees, no subscriptions, and no credit card required. Heoster's mission is to democratize AI access for every student and developer.

Question 3

Do I need to create an account?

Accepted Answer

No account is required for basic usage. Creating a free account lets you save conversation history across devices, sync settings, and access personalized features. Sign up is optional and always free.

Question 4

What can I ask SOHAM?

Accepted Answer

Anything: coding help, math problems, general knowledge, homework, document analysis, image generation, web searches, creative writing, debugging, architecture advice, translations, summaries, and more. The AI adapts to your query automatically using smart routing.

Question 5

What does "SOHAM" mean?

Accepted Answer

SOHAM has two meanings. In the product, it stands for "Self Organising Hyper Adaptive Machine." The name is also inspired by the Sanskrit mantra Soham (So Hum), meaning "I am That" — representing unity of individual self with universal consciousness. Heoster chose this name to give the product both technical identity and human depth.

Question 6

Who built SOHAM?

Accepted Answer

SOHAM was built by Heoster (Harsh), a 16-year-old developer from Khatauli, India. He founded CODEEX-AI in 2024 and built SOHAM as the flagship product. You can learn more on the About page.

Question 7

How many AI models are available?

Accepted Answer

SOHAM provides access to 35+ AI models from Groq (Llama 3.1 8B Instant, Llama 3.3 70B, Mixtral 8x7B), Google (Gemini 2.5 Flash, Gemini Flash Latest), Cerebras (Qwen 3 235B Instruct, GLM 4.7), HuggingFace (DeepSeek R1, GPT-OSS, RNJ-1), and OpenRouter (Auto free tier).

Question 8

What is Auto mode?

Accepted Answer

Auto mode is SOHAM's smart routing system. It analyzes your query and automatically selects the best model — math questions go to math-capable models, coding questions to DeepSeek or Llama 3.3, image tasks to Gemini, and so on. It's recommended for most users.

Question 9

Which models are completely free with no limits?

Accepted Answer

HuggingFace models (DeepSeek R1, GPT-OSS, RNJ-1) and OpenRouter Auto free tier have no usage limits. These are marked "Free" in the model selector.

Question 10

Which model is fastest?

Accepted Answer

Groq's Llama 3.1 8B Instant is the fastest — Groq's hardware is purpose-built for ultra-low latency inference. Use it when you need quick answers and speed matters more than depth.

Question 11

Which model is best for coding?

Accepted Answer

DeepSeek R1 (HuggingFace) is the best coding model available in SOHAM. For general programming questions, Llama 3.3 70B also performs very well. Auto mode will route coding questions to these models automatically.

Question 12

Which model is best for long documents?

Accepted Answer

Google Gemini 2.5 Flash has the largest context window and is best for long documents, PDFs, and complex multi-step reasoning tasks.

Question 13

Can I switch models mid-conversation?

Accepted Answer

Yes. Open the model selector in the top bar and switch at any time. The conversation history is preserved and passed to the new model.

Question 14

What are slash commands?

Accepted Answer

Slash commands are shortcuts that route your query to a specialized pipeline. /solve routes to math/coding models with step-by-step output. /search triggers DuckDuckGo web search. /summarize condenses long text into key points. /translate translates text. /classify categorizes content. You can also just ask naturally — SOHAM understands intent.

Question 15

Does web search trigger automatically?

Accepted Answer

Yes. SOHAM's intent detector scans every message for signals that need real-time data — keywords like "today", "latest", "current", "news", "price", "who is the current...", etc. When detected, it automatically fetches DuckDuckGo results and synthesizes them with source links.

Question 16

How does voice input work?

Accepted Answer

Click the microphone icon next to the message input. Allow microphone access when prompted. Speak clearly — Groq Whisper V3 Turbo transcribes your speech in real time. The transcription appears in the input field for you to review before sending.

Question 17

How does text-to-speech work?

Accepted Answer

Enable speech output in settings and choose a voice. You can also click the speaker icon on any assistant message to hear it read aloud. SOHAM uses Groq's Orpheus TTS with a fallback chain to Edge TTS and browser TTS.

Question 18

How does image generation work?

Accepted Answer

Type "generate an image of..." or "create a picture of..." in chat. SOHAM routes this to the image generation pipeline using HuggingFace FLUX.1-schnell. The generated image appears inline in the chat with download and preview options.

Question 19

How does PDF analysis work?

Accepted Answer

Go to /ai-services or /pdf-analyzer. Upload a PDF (up to 5 MB). Ask questions about the document — SOHAM extracts text, analyzes content, and answers your questions. Great for research papers, contracts, reports, and documentation.

Question 20

Does SOHAM remember our conversation?

Accepted Answer

Yes, within a session. The AI maintains full context across all messages in a conversation. You can reference earlier messages naturally: "explain that differently" or "give me another example of what you showed earlier."

Question 21

Can I export my conversations?

Accepted Answer

Yes. Use the export button in the chat sidebar to download conversations as JSON. You can also copy individual messages or use the share feature.

Question 22

The AI isn't responding or is very slow

Accepted Answer

Try: (1) Refresh the page. (2) Check your internet connection. (3) Switch to a different model in settings — free HuggingFace models may be slower during peak times. (4) Try a Groq model for fastest response. (5) Clear browser cache if the issue persists.

Question 23

Voice input isn't working

Accepted Answer

Check that your browser has microphone permissions (click the lock icon in the address bar). Make sure your microphone works in other apps. Voice input requires HTTPS. Try Chrome or Edge for best compatibility. Disable browser extensions that might block microphone access.

Question 24

I can't install the mobile app

Accepted Answer

Android: Use Chrome, visit the site, tap the menu (⋮), and select "Add to Home screen." iOS: Use Safari only (not Chrome), tap the share button (□↑), then "Add to Home Screen." Make sure you're on the latest browser version.

Question 25

My settings aren't saving

Accepted Answer

Settings are stored in browser local storage. Make sure cookies and local storage are enabled. Settings won't persist in private/incognito mode. Try clearing browser cache and reconfiguring.

Question 26

The AI gave an incorrect answer

Accepted Answer

Try: (1) Rephrase with more specific details. (2) Use /solve for math/coding, /search for current info. (3) Switch to a different model manually. (4) For current events, make sure web search triggered (you'll see source links in the response). (5) Ask the AI to "try again" or "reconsider."

Question 27

Image generation failed

Accepted Answer

HuggingFace free tier has rate limits. Wait a few minutes and try again. Make sure your prompt is clear and descriptive. Avoid prompts that might violate content policies. Try simplifying the prompt if it's very complex.

Question 28

The page is loading slowly

Accepted Answer

SOHAM is a PWA — after the first load, it caches resources for faster subsequent loads. If it's consistently slow, try: clearing cache, disabling browser extensions, or switching to a faster network.

Question 29

How do I install SOHAM on my phone?

Accepted Answer

SOHAM is a Progressive Web App (PWA). Android: Open in Chrome → menu (⋮) → "Add to Home screen." iOS: Open in Safari → share button (□↑) → "Add to Home Screen." The app will work like a native app with its own icon.

Question 30

Does the installed app work offline?

Accepted Answer

The app shell loads offline (you can open it without internet). However, generating new AI responses requires an internet connection. Previously loaded conversations are accessible offline.

Question 31

Why install instead of using the browser?

Accepted Answer

The installed PWA gives you: full-screen experience without browser UI, faster loading with cached resources, home screen icon for quick access, and better integration with your device's app switcher.

Question 32

Is there a native Android or iOS app?

Accepted Answer

Not yet — the PWA provides a near-native experience on both platforms. A native app is on the roadmap for future development.

Question 33

Is my data safe?

Accepted Answer

Yes. Conversations are stored locally in your browser's local storage — not on SOHAM servers. We use HTTPS for all connections. We don't log or monitor your conversations. Your privacy is a core design principle.

Question 34

Do you use my conversations to train AI models?

Accepted Answer

No. Your conversations are not used to train AI models and are not shared with model providers for training purposes.

Question 35

Can I delete my conversation history?

Accepted Answer

Yes. Clear conversation history anytime through the settings panel or by clearing your browser's local storage for soham-ai.vercel.app.

Question 36

Is SOHAM GDPR compliant?

Accepted Answer

Yes. We follow GDPR principles: data minimization, user control, transparency, and no data selling. Read our full Privacy Policy for details.

Question 37

What data does SOHAM collect?

Accepted Answer

Minimal data: anonymous usage analytics (page views, feature usage) via Google Analytics. No conversation content, no personal data without explicit consent. Account users store only email and display name for authentication.

Question 38

How can I contribute to SOHAM?

Accepted Answer

Visit our GitHub repository to open issues, suggest features, or submit pull requests. Community feedback directly shapes the product roadmap.

Question 39

Is there a community forum or discussion board?

Accepted Answer

Yes! Visit the /community page to share posts, ask questions, and interact with other SOHAM users. Posts are stored publicly and help build a knowledge base for everyone.

Question 40

How can I report a bug?

Accepted Answer

Use the contact form at /contact, email us directly, or open a GitHub issue. Include: what you were doing, what happened, what you expected, your browser and OS, and any error messages.

Question 41

Can I suggest new AI models to add?

Accepted Answer

Absolutely. Use the contact form or GitHub issues to suggest models. We evaluate new models based on quality, availability, cost, and user demand.

Everything you need to know about SOHAM

Jump to a Section

Getting Started

AI Models & Selection

Features & Commands

Technical Issues & Troubleshooting

Mobile & PWA

Privacy & Security

Community & Contributions

Explore SOHAM Features

Still Need Help?

Contact Support

Documentation

Community