An open-source research agentest. 2024
Scira reads the
web for you,
honestly.
It searches deep, cites every claim, and connects to the apps you already use, so the work doesn't end at the answer.
- Searches
- 5M+
- Users
- 100K+
- Stars
- —+
Most search tools answer fast and forget what they said. Scira treats research like a craft. It plans, it retrieves, it cross-checks. Every claim links back to a source you can open and read.
It's open source under AGPL-3.0. Self-host it, fork it, point it at your own models. The whole system is yours to inspect.
“Better than Grok at digging up information from X. Insanely accurate answers.”
Four ideas hold the whole thing together. They're boring in isolation, interesting in combination.
- 01
Plan before you fetch
A complex question becomes a list of sub-questions. Each one gets the right model and the right tool, run in parallel where possible.
- 02
Cite everything
No claim without a source. Inline citations let you click through to the exact page the agent read. If the source disagrees with the answer, you find out fast.
- 03
Stay open
AGPL-3.0 source. Bring your own keys. Self-host. Plug in custom tools. The agent should bend to your workflow, not the other way around.
- 04
Run on a schedule
Lookouts watch a topic for you on a schedule. Daily, weekly, whatever rhythm fits. They check back, notice changes, and email what shifted.
One input box. 17 ways to point it. Each mode is tuned for a different kind of source. Pick one, or let Scira choose.
13 free · 4 pro
“Does a really good job scraping through the reddit mines.”
Connect 100+ apps via MCP and Scira can act inside them. Open a PR. Post to Slack. Update a Notion page. Research that ends with a button click instead of a copy-paste.
16 of 100+ shown
Browse all apps“Searched for myself in extreme mode. Not just the best, it's wild. And 100% accurate.”
126 models from every major provider, switchable mid-conversation. Use whichever one fits the question.
Grok 4.1 Fast
xAI's greatest and fastest multimodel LLM
Qwen 3 32B
Alibaba's base LLM
Qwen 3 32B Thinking
Alibaba's base reasoning LLM
Nemotron 3 Super
NVIDIA's powerful Nemotron 3 model
Qwen 3 4B
Alibaba's small base LLM
Qwen 3 4B Thinking
Alibaba's small base LLM
Gemini 2.5 Flash Lite
Google's advanced small LLM
Ministral 3 3B
Mistral's mini-model 3B multi-modal LLM
Ministral 3 8B
Mistral's mini-model 8B multi-modal LLM
Devstral 2
Mistral's coding-focused LLM
Devstral Small 2
Mistral's small coding-focused LLM
Mistral Small 4
Mistral's small efficient model
Free is good for casual use. Pro is for serious research. Max adds the full Anthropic Claude lineup, Gemini 3.1 Pro, and the Scira Agent preview. Handy for long-form work.
| Plan | Free$0 /mo | Pro$15 /mo | Max$60 /mo |
|---|---|---|---|
| Research runs | 3/day | Unlimited | Unlimited |
| Search modes | 13 | All 18 | All 18 |
| Base AI models | Basic | All | All |
| Claude Haiku 4.5 | Unlimited | In 120/wk pool | |
| Claude Sonnet 4.6 · Opus 4.7 | 150 · 30 msgs/mo | 120/wk pool | |
| OpenAI GPT 5.4 · 5.5 (pooled) | 40/wk | 40/wk | |
| Gemini 3.1 Pro | 80/mo | ||
| PDF analysis | |||
| Voice mode | |||
| Connectors (Drive, Notion) | |||
| Lookout · Memory · XQL | |||
| Canvas | |||
| Scira Agent (preview) | |||
| Priority support | |||
“Read nothing the whole sem and topped my mid-sems with Scira. Diagrams, points, topics, all there.”
What is Scira?+
An open-source AI assistant built for research and action. Searches the web in depth, cites its sources, and connects to 100+ apps via MCP so you can act on what you find.
Free vs Pro vs Max?+
Free gives you 3 runs/day with basic models. Pro ($15/month) unlocks unlimited research, all base models, Claude Haiku unlimited, Sonnet 4.6 (150 msgs/mo), Opus 4.7 (30 msgs/mo), PDF, voice, Lookouts. Max ($60/month) adds the full Claude pool (120/wk), Gemini 3.1 Pro (80/mo), and Scira Agent.
How are model quotas enforced?+
Pro caps Claude Sonnet 4.6 at 150 messages and Opus 4.7 at 30 messages per calendar month. Max pools all Anthropic Claude requests at 120 per calendar week. Both plans share a weekly pool of 40 OpenAI GPT 5.4 / 5.5 requests. Limits reset automatically and are visible in your account.
Can I cancel anytime?+
Yes. Benefits stay until the end of the billing period.
Which AI models does Scira use?+
Grok, Claude, GPT, Gemini, and others. Switch per query.
How does Scira ensure accuracy?+
Outputs are grounded in retrieved sources with inline citations. Agents cross-check before synthesizing.
Anything else? Email zaid@scira.ai.
- Inference
- Vercel AI SDK
- Search
- Exa
- Voice
- ElevenLabs
- Hosting
- Vercel
- Payments
- Polar · DodoPayments
- License
- AGPL-3.0