Run open models and stop paying frontier prices
Run open models whenever you can, on your own hardware or a host like z.ai. Bring in a frontier model when a job calls for it, and stop paying frontier rates for work that doesn't need them.
Run open models whenever you can. Frontier models are useful, and some jobs still want the strongest model you have. But lint fixes, test runs, migrations, doc passes, and a growing share of real coding work run fine on an open model on your own hardware, or through a managed host like z.ai. Every job you move over saves premium tokens, and it puts weight behind models that get better the more they're used.
Setup happens once. Point crystl at Ollama, LM Studio, vLLM, llama.cpp, or a compatible hosted service in Settings → agents → providers & keys. crystl checks the model is actually reachable, keeps your key in the Keychain, and hands every new shard the right flags and environment variables so you never have to remember them.
Put the frontier model in charge, not on every task. Orchestration is where a frontier model is still the smart spend. A strong lead plans the work, hands bounded pieces to open-model workers, and judges what comes back. Each worker is still a normal shard: you can watch it, steer it, and pull its result back into the bigger job.
Open models are not automatically private, and they are not all equally capable. Model and agent both on your own hardware? The work stays local. A managed host still sees your code, and a local worker under a hosted lead is not an air gap. Pick the model for the task, and the price takes care of itself.
Why it matters
Set up once
Connect a local runtime or hosted service one time. crystl handles the keys, the agent-specific flags, and the environment for every shard after that.
Orchestrate every model
Mix open-model and frontier workers in one job. Open models take everything they can, a frontier model takes what's left, and you decide where the line is.
Cut token costs
Move the repetitive, high-volume work off metered frontier APIs. Save the expensive tokens for decisions.
Know what is ready
See which connections are up and which Ollama models are warm before you launch. A model that is down fails at setup, not twenty minutes into a job.
Three steps from model to worker
Start the model, name the connection, pick it when you open a shard. crystl does the agent-specific plumbing. Your key stays in the Keychain, and anything you export yourself still wins.
Settings → agents → providers & keys → open models
label closet-box
kind Ollama
base URL http://100.x.y.z:11434 ← as seen from your Mac
as seen from remote http://localhost:11434 ← for agents running on the box
key (none)
env vars OLLAMA_API_BASE, OLLAMA_HOST
enabled yes
label z-ai
kind anthropic-compatible
base URL https://api.z.ai/api/anthropic
key •••••••••••• ← Keychain, never your Anthropic key
env vars ANTHROPIC_BASE_URL, ANTHROPIC_AUTH_TOKEN
enabled yes
New shards pick this up at launch; a shard that was already running keeps its old values until you restart it. Full walkthrough in the open models setup guide.
Orchestrate by model size
Every agent spells model ids its own way. Skip the lookup: crystl reads what your connection serves and proposes a small, standard, and large model. Approve the mapping once, and heroes, quests, and orchestrators ask for a size instead of an id.
$ crystl agent profile suggest --agent codex --slot closet-box
closet-box serves 6 models. proposed profile for codex:
small gpt-oss:20b
standard qwen3-coder:30b
large qwen3-coder:480b
write this profile? [y/N]
aider ollama_chat/qwen3-coder # ollama_chat beats ollama here
opencode ollama/qwen3-coder
That one map makes model choice an orchestration control. Cheap workers run small or standard, large is saved for the hard stuff, and you can swap the underlying model later without touching a single hero or workflow.
Where open models pay off
High-volume mechanical work
Lint fixes, test loops, migrations, doc passes: bounded work an open model chews through happily, with no premium tokens involved.
A mixed-model agent team
A strong lead plans the work and hands well-scoped pieces to open-model workers. Every worker stays visible, and every result has a clear path back.
Your own hardware
Run agent and model on a Mac, workstation, or GPU box you control. The remote-vantage URL keeps the same model reachable from your desk and from an SSH session on the box itself.
Hosted open models
No GPU? A managed host like z.ai runs the model for you, through the same shards, model sizes, and orchestration controls.
Setup tutorials
Each one goes from nothing installed to a working shard on that server.
Ready to try it?
Create a free account and download crystl for macOS.
get startedgetting started guide →Frequently asked questions
Crystl's Book of Spells
Master crystl with a grimoire of powerful prompts. Summon heroes, assemble a party, revive shards, and complete epic quests.
open the book →More features
Gems & Shards
Gems, shards, and the crystal rail keep every project one click away.
Action Panels
See what every agent is doing at a glance.
Agent Activity Panel
One live view of every agent across all your gems: who's working, who's idle, and where tokens are going.
Schedule Agents
Launch a fresh agent at a set time (once, daily, or weekdays). Pick the gem, agent, model, and prompt; crystl fires it on schedule.
Automation (Sequences)
Automate the work you keep doing by hand. Write the job down once and let a schedule, a file landing, or a watched URL run it for you.
Parallel Sessions
Run multiple Claude instances on the same repo in separate git worktrees.
Hero Shards
Summon a single hero as a shard (its own agent, model, and persona) without a quest party.
History Navigator
Browse, jump through, and search every session: conversation and tool calls.
Token Tracking
Track token burn per shard, estimate turns left, and reclaim context window by disabling heavy plugins and skills.
Workbench
A slide-out task list in every project, backed by a WORKBENCH.md the agent reads and writes with you.
Markdown Editor
Render and edit any .md in a side panel. Toggle it from the workbench, and let agents open docs for you from the CLI.
Prompt Integration
Facet Inserts give you one-click access to saved prompts, commands, and shortcuts.
Agent File Editor
Create, edit, and manage CLAUDE.md, rules, skills, and agent config files with a block editor, file chooser, and project optimizer.
Project Optimizer
Scan your project for gaps in agent configuration (missing files, incomplete docs, oversized code, and setup issues) with one-click fixes.
Skills Library
Browse and import agentic skills that teach agents like Claude Code or Codex new tricks: code review, testing, deployment, and more.
Cavrn
The terminal as structured data: every turn, tool call, result, and table captured as agent-readable metadata. Drawn fast on the GPU with Metal.
Remote Development
Full crystl integration over SSH: approval panels, file relay, and click-to-open on remote machines.
Agent Orchestration
One agent spawns, supervises, and merges the work of many: a control room for a team of coding agents.
crystl quest
Assemble a party of specialized agents. Each brings unique skills, coordinates in shared chat, and executes in parallel.
Screenshot Bar
Drag recent screenshots into the terminal. Spotlight-powered, always up to date.
Copy Bar
A one-click copy bar under the terminal. Agents stage text, you click Copy. Editable, up to 10 tabs, free.