Cumulus LabsBack to Leaderboard
LLMs
Qwen 3 8B
+22.4%~$QW3-8
High-volume 8B instruction model with thinking mode. Most deployed model on Ion by request count. Sub-second TTFB on GH200. Runs on every universal and video node.
Deploy Now
#1 overall · 6 deploys
1.2s
Cold Start
avg on Ion
203ms
Avg Inference
per request
6
Active Replicas
right now
6
Total Deployments
all time
7-Day Trend
+22.4% this weekCold Start vs Competitors
Lower is better●~$QW3-81.2s
~$QW301.7s
~$QW321.8s
~$QW141.4s
Cold start comparison vs similar models. Lower is better.
Cost Estimate
Cost per 1K Inferences
$0.45
Est. Daily (1K req/day)
$0.45
Est. Monthly (30K req)
$13.50
No subscriptions. Buy credits, pay per inference. Scale to zero when idle.
Quick Deploy
cumulus-sdkimport cumulus from "cumulus-sdk" // Deploy Qwen 3 8B on Ion const client = await cumulus.deploy("qwen3-8b") // Run inference const result = await client.run({ prompt: "Your prompt here", // model-specific params... })