Meterless
SWARMS · V0.9.3 · DAG
SWARMS, THE MANY. · FREE & OPEN SOURCE

The swarm
builds itself.
You keep it.

Type one goal. Swarms auto-generates the whole agent team: specialized agents planned, parallelized, reconciled, and verified, running locally in your browser. The reasoning is yours to keep: edit the graph, rerun any branch, refine it over time.

View on GitHub
Swarm size300+ agents
Context costO(1)
Providers5 / 2 local
Swarm designAuto-generated
RunsLocally · in-browser
PriceFree · open source
SWARM · batch_9f2a · running
done0/50 tokens0 wall0.0s
PROMPT Can you analyze this product overview of a Meterless application called Gaia, and generate 30 variants of a interactive engaging loading screen for it. EDIT

DAGs at scale running locally

The answer is one asset. The graph that produced it is the bigger one. Use natural language to create and refine reasoning topologies for research, comparison, coding, analysis, writing, batch work, and verification, with local control by default, scheduled to run automatically.

PROMPT · your input
$ swarm
GENERAL
try:
DAG · idle
parsed intent
spawn plan
est. nodes
awaiting input.
Nodes
147
Edges
382
Depth
14
Critics
22
Merges
19
node -
SEED
initial_promptroot
modelhaiku-4.5
tokens412
latency182ms
score0.91
drag to pan · scroll to zoom · click any node · space to traverse

Stop streaming AI.
Start owning it.

Other AI tools make you pay to re-run the model every single time. Swarms saves the reasoning as a graph you own and can run again.

EVERY OTHER AISTREAMING
RUN 01
full price
RUN 02
full price
RUN 03
full price
YOU KEEP · NOTHING
SWARMSYOUR LIBRARY
RUN 01 · frontier
RUN 02 · cheap
RUN 03 · local
YOU KEEP · GRAPH · ARTIFACTS · LINEAGE
01 · INSPECTOpen any nodemodel · tokens · score 02 · EDITRefine the graphselect · describe · diff · apply 03 · SWAPAny model, any nodefrontier for hard · cheap for the rest 04 · RERUNRun it againhistory · branch · schedule

Frontier once. Then cheaper models riding a refined graph. The work compounds.

Swarm math.

Their swarm runs on their cloud. With Meterless yours runs on your machine at a fraction of the token cost per completed agentic task.

0
Scout intent modes
0ms
Intent classification
+
Model providers
+
Context-sharing agent swarms
END · SWARMS

Stop generating one at a time.
Swarm the problem.

The swarm is auto-generated, runs on your machine, and every graph it builds stays yours to edit, rerun, and refine. Free and open source. Need the project, memory, and results to persist across sessions? Continue the work in Gaia.

Explore Gaia →