OpenJar

OpenJar

5 AI minds. One team. They remember everything.

1

Bring your problem

Chat, API, or full pipeline. However you work, the team's ready.

2

The team goes to work

5 models from 4 companies — each with a role, a personality, and access to every solution they've ever seen.

3

Smarter every time

Solutions get stored. Models critique each other. The team learns from every jar it opens.

🧠 The Team

Five minds from four companies. They know each other. They challenge each other. They remember.

🟣
Claude Opus 4.6
The Architect
Anthropic
Sees the big picture. Breaks hard problems down. Resolves when the team disagrees.
🔵
Claude Sonnet 4.6
The Editor
Anthropic
Nothing gets past her. Catches edge cases, finds the nuance everyone else missed.
🟢
GPT-5.2
The Engineer
OpenAI
Builds things. Code, data, structured output. The Architect designs — he constructs.
⚫
Grok 4
The Wildcard
xAI
Challenges everything. Asks "what if we're wrong?" Sometimes way off. Sometimes the only one who's right.
🔴
Gemini 2.5 Pro
The Researcher
Google
Goes deep. Long-context reasoning, citations, multi-step analysis. Gets the hard research questions.
Memory
→
Decompose
→
Specialists
→
Cross-Check
→
Resolve
→
Retrospective
→
Learn

🫙 Open a jar

The hard problem. The stuck project. The thing nobody else could crack.

Estimated: ~0 🥒

💬 Team Chat

Talk to any team member directly, or let them all weigh in. They share memory — what one learns, they all know.

🫙 Opening...

Phase 0 — Memory Search
🧠 Searching past solutions Waiting
Phase 1 — Decompose
🟣 The Architectbreaking it down Waiting
Phase 2 — Specialist Work
🟣 The ArchitectAnthropic Waiting
🔵 The EditorAnthropic Waiting
🟢 The EngineerOpenAI Waiting
⚫ The WildcardxAI Waiting
🔴 The ResearcherGoogle Waiting
Phase 3 — Cross-Check
🔍 Each answer verified by a different model Waiting
Phase 4 — Resolve & Assemble
🎯 The Architectfinal synthesis Waiting
Phase 5 — Retrospective & Learn
🪞 All models critique the answer Waiting
🔧 Refactorif improvements found Waiting
📚 Store in team memory Waiting

🫙 Jar Opened

🥒 Pickles

Pickles fuel the team. No subscriptions — buy what you need, use it when you want.

5K 🥒
$10
$0.002 / pickle
25K 🥒
$40
$0.0016 / pickle
100K 🥒
$120
$0.0012 / pickle

📡 API

Same team, same memory. Build it into your apps, CI, workflows — wherever you need the room.

POST /v1/auth/signup

Create an account. Returns your API key + 500 free pickles.

curl -X POST https://api-openjar.allautotunes.com/v1/auth/signup \
  -H "Content-Type: application/json" \
  -d '{"email": "you@example.com", "password": "your-password"}'

# Response:
# { "api_key": "oj_sk_a1b2c3...", "balance": 500 }

POST /v1/auth/login

Log in to get your API key.

curl -X POST https://api-openjar.allautotunes.com/v1/auth/login \
  -H "Content-Type: application/json" \
  -d '{"email": "you@example.com", "password": "your-password"}'

POST /v1/solve

Full pipeline solve. Memory search → decompose → specialists → cross-check → resolve → retrospective → learn.

curl -X POST https://api-openjar.allautotunes.com/v1/solve \
  -H "Authorization: Bearer oj_sk_your_key" \
  -H "Content-Type: application/json" \
  -d '{"problem": "How do I architect a distributed cache that handles 1M req/s?"}'

POST /v1/chat

Conversational mode. Pick a model or set team_discuss: true for all five.

curl -X POST https://api-openjar.allautotunes.com/v1/chat \
  -H "Authorization: Bearer oj_sk_your_key" \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [{"role": "user", "content": "Explain CRDT vs OT"}],
    "model": "grok",
    "team_discuss": false
  }'

POST /v1/balance

Check your pickle balance and available packs.

curl https://api-openjar.allautotunes.com/v1/balance \
  -H "Authorization: Bearer oj_sk_your_key"

POST /v1/memory/search

Search the team's shared knowledge from past solves.

curl -X POST https://api-openjar.allautotunes.com/v1/memory/search \
  -H "Authorization: Bearer oj_sk_your_key" \
  -H "Content-Type: application/json" \
  -d '{"query": "distributed systems consensus", "limit": 5}'