Now in public beta
Conversations that never lose the thread.
Longthread gives your AI applications infinite conversation memory through intelligent context compaction. Automatic, provider-agnostic, drop-in compatible.
$ curl https://api.llm.archi/v1/chat/completions \
-H "Authorization: Bearer ${LONGTHREAD_API_KEY}" \
-d '{"model": "gpt-4o", "messages": [...]}'How it works
Four layers of context,
one focused window.
Longthread automatically manages your conversation context through multi-level compaction. Your model always gets the most relevant information within its token window.
System Prompt
Your instructions and persona definition — always present, never compacted.
Section Summaries
Completed conversation sections are distilled into structured summaries — capturing key facts, decisions, and open questions.
Ledger Paragraphs
Older turns compressed into concise ledger entries — the essential facts without the filler.
Recent Raw Turns
The latest messages preserved verbatim — full nuance, full context, no summarization.
Features
Built for production
AI applications.
Zero-config compaction
Automatic, intelligent context management. No manual prompts, no chunking logic, no context window math. Just send messages.
Drop-in API
OpenAI-compatible REST API. Swap your base URL and add your API key. Existing SDKs work out of the box.
Provider-agnostic
Route to OpenAI, Anthropic, or any compatible provider. Switch models without changing your integration.
Context preservation
Facts, decisions, user intent, and open questions survive compaction. Your AI remembers what matters.
Metered billing
Pay only for the tokens you use. No per-seat charges, no upfront commitments. Start free, scale as you grow.
Edge-native
Built on Cloudflare Workers. Sub-millisecond cold starts, global distribution, automatically scales to zero.
Pricing
Start free.
Scale when ready.
Simple, transparent pricing based on the tokens you actually use.
Free
For prototyping and side projects
- 100K tokens/month
- All providers
- Level 1 compaction
- Community support
Pro
For production applications
- 5M tokens/month included
- All providers + priority routing
- Level 1 & 2 compaction
- Email support
- Usage dashboard
Enterprise
For teams with custom needs
- Unlimited tokens
- Dedicated infrastructure
- All compaction levels
- SLA guarantee
- Custom integrations
- Dedicated support
Ready to go infinite?
Get your API key in 30 seconds. No credit card required.
# Install the SDK
npm install @longthread/sdk
# Or just use curl
export LONGTHREAD_API_KEY="lt_..."
curl https://api.llm.archi/v1/chat/completions \
-H "Authorization: Bearer $LONGTHREAD_API_KEY"