netoperationalslot297,841,022nodes412 / 487utilization68%p5014.2 ms/tok
◇llama-3.1-70b24.1ms/tok◇qwen-2.5-72b18.7ms/tok◇mixtral-8x22b31.4ms/tok◇deepseek-r152.3ms/tok◇phi-48.1ms/tok◇command-r-plus22.9ms/tok◇gemma-2-27b14.6ms/tok◇llama-3.1-70b24.1ms/tok◇qwen-2.5-72b18.7ms/tok◇mixtral-8x22b31.4ms/tok◇deepseek-r152.3ms/tok◇phi-48.1ms/tok◇command-r-plus22.9ms/tok◇gemma-2-27b14.6ms/tok◇llama-3.1-70b24.1ms/tok◇qwen-2.5-72b18.7ms/tok◇mixtral-8x22b31.4ms/tok◇deepseek-r152.3ms/tok◇phi-48.1ms/tok◇command-r-plus22.9ms/tok◇gemma-2-27b14.6ms/tok
buildv0.7.4-rc.2
▲ NnimbutRC.2
Try inferenceRun a node →
◇ catalog · v0.7.4-rc.247 models total · 9 featured below

The catalog.

Open-source frontier models, served from 412 nodes. No closed weights, no API keys, no platform invoices. Pick a model, sign a request, pay per token.

◇ filterAllGeneralReasoningTool-useSmallMoE
◇ meta70B · 128k

llama-3.1-70b-instruct

general · stable · most-served
p50 ms/tok
24.1
ttft
142ms
cost / 1k tok
$0.00040
Try →
◇ alibaba72B · 128k

qwen-2.5-72b-instruct

general · fast · cheap
p50 ms/tok
18.7
ttft
98ms
cost / 1k tok
$0.00038
Try →
◇ deepseek671B · 64k

deepseek-r1

reasoning · slow · premium
p50 ms/tok
52.3
ttft
412ms
cost / 1k tok
$0.00210
Try →
◇ mistral141B · 64k

mixtral-8x22b

mixture-of-experts
p50 ms/tok
31.4
ttft
188ms
cost / 1k tok
$0.00064
Try →
◇ microsoft14B · 32k

phi-4

small · cheap · fast
p50 ms/tok
8.1
ttft
44ms
cost / 1k tok
$0.00018
Try →
◇ cohere104B · 128k

command-r-plus

tool-use specialized
p50 ms/tok
22.9
ttft
156ms
cost / 1k tok
$0.00050
Try →
◇ google27B · 32k

gemma-2-27b

on-the-edge ready
p50 ms/tok
14.6
ttft
72ms
cost / 1k tok
$0.00024
Try →
◇ zhipu9B · 128k

glm-4-9b

tiny · multilingual
p50 ms/tok
6.4
ttft
38ms
cost / 1k tok
$0.00012
Try →
◇ nvidia340B · 32k

nemotron-340b

synthetic-data king
p50 ms/tok
62.1
ttft
488ms
cost / 1k tok
$0.00184
Try →

◇ ranked · by cost-per-quality

Side-by-side, head-to-head.

Quality is the GPQA-Diamond benchmark, retested weekly. Cost is what you actually pay per million tokens at current spot rates. We sort by their ratio.

modelparamsgpqa-d$/1m tok$/quality ptactions
qwen-2.5-72b72B54.2$0.38$0.0070testdocs
deepseek-r1671B71.8$2.10$0.0292testdocs
llama-3.1-70b70B50.1$0.40$0.0079testdocs
gemma-2-27b27B42.4$0.24$0.0056testdocs
phi-414B38.8$0.18$0.0046testdocs
command-r-plus104B47.2$0.50$0.0105testdocs
mixtral-8x22b141B49.6$0.64$0.0129testdocs

Pick a model.
Send a prompt.

The interactive playground is one click away. No signup, no key — just type.

Open the playground →Add a model to the catalog