Skip to the document
Madhuopen lab

AI Insights: Latency & Cost Optimization · part 11 of 17

insight27 July 2026cost: ~$4.20 (36-min bench, auto-teardown)

Foundry managed compute: Qwen without running the VM

Same model from the Azure Foundry catalog on managed H100, benchmarked end to end

The question

Can the winning Qwen recipe run on Azure Foundry managed compute — no VM, no vLLM ops — and what does that cost?

insight

Managed H100 is ~15% costlier per answer than the DIY A100 but 1.6x faster wall time — a workable option that removes VM ops.

Wall (single, V5)5.5 s1.6x faster than DIY A100
TTFT544 msgateway overhead; DIY = 92 ms
Cost/answer$0.00149@ $6.98/hr, conc 16
Throughput sweep660–4,680 /hr@ conc 1/4/8/16

Results (n=10, V5 prompt, H100 BF16)

Results (n=10, V5 prompt, H100 BF16)
MetricFoundry H100 (managed)DIY A100 + vLLMGPT-5.1 API
Wall time (single)5.5 s9.0 s15–24 s
TTFT544 ms (gateway)92 ms~1,500 ms
Quality (Opus/Sonnet)7.0 / 6.07.25 / 7.0 (FP8)8.0 / 7.5
Ungrounded numbers (median)1.5 (BF16 regression)0 (FP8)0
Cost per answer$0.00149$0.00127$0.0136
180M-call season~$268K~$229K~$2.46M

Deployment mechanics that cost a day

  • The FP8 H100 template is dead — deployed 'Succeeded' twice but returned 500 forever. The BF16 1xH100 template works on first ping.
  • Auth is the api-key header — Bearer + account key returns 401.
  • The model field in requests must be the deployment name, not the model id.
  • Deployment templates are listed via azure-ai-ml MLClient(registry_name=...).deployment_templates.list() with DeploymentTemplate._from_rest_object monkeypatched (SDK parse bug); deployment itself via azure-mgmt-cognitiveservices==15.0.0b2 managed_compute_deployments.
  • Direct HTTPS endpoint, no tunnel: .../managed-deployments/<name>/v1/chat/completions (also /v1/messages). Managed-compute quota is separate from VM quota — 1xH100 fit the existing allowance.
Azure Foundrymanaged computeH100deployment