The 1-day diagnostic.
You tell us the model, the config, and the number you're stuck at. For models and frameworks already on our lab, one working day after payment and materials are in: why it's slow, or why it can't be reproduced — on our own multi-node GB10 lab, in writing, with the evidence. Otherwise we give you the date before you pay. Then you decide what happens next.
What you get
Reproduction on real hardware
We stand up your configuration — framework, model class, quantization, topology — on our own standing GB10 cluster. No access to your production systems required. If it can't be reproduced, that finding is itself diagnostic, and we say so.
A written root-cause report
Which of the known failure classes you're in, the evidence that puts you there, what it's costing you in tokens/sec or engineer-days, and the recommended fix — scoped, with an honest estimate of the work.
Measured numbers
Every claim in the report is backed by paired measurements of delivered tokens on identical hardware, with confidence intervals wherever the quantity is a throughput we sample repeatedly — the same discipline as our published field notes. We state the protocol so you can re-run it, and restart variance is reported, not hidden.
See the format first: a sample report, written from a case on our own cluster, in the same format as a client report. Common questions are answered in the FAQ.
How it runs
Simple on purpose. You never float our costs; we never waste your day.
- 1 · You write. Model, framework + version, node count, interconnect, the symptom, and the number you're stuck at. For the fit check, model, versions, node count and the symptom are enough; send configs and logs after the authorization is signed.
- 2 · We confirm fit — honestly. If it's not a fit (wrong hardware, wrong problem class), we tell you straight and point you somewhere useful. No charge for finding out.
- 3 · Sign one form. The client authorization covers confidentiality both ways, and access and recording if the work ever needs your systems. Sign it by e-signature or return a signed PDF by email. Read the sample now.
- 4 · Invoice, paid up front. Fixed $1,500. The clock starts when payment lands and materials are in.
- 5 · One working day later (for models and frameworks already on our lab; otherwise we give you the date before you pay): the written report. If the fix is a one-liner, it's in the report — done, no upsell. If it's real work, the report ends with a scoped fixed bid or hourly estimate. Taking it is optional.
The guarantee: if the report isn't actionable, we refund it in full. Actionable means at least one measured finding — a confirmed root cause or quantified recoverable headroom — not just generic advice. If the report doesn't contain that, we refund the full fee promptly, by the method you paid. We can offer this because a slow GB10 cluster almost always has recoverable headroom once you know where this platform hides it.
Is your problem in scope?
If it's on the trap index, yes — that list is literally our diagnostic map. Multi-node bring-up, throughput and latency, weight-load stalls, unified-memory behavior, speculative decoding, quantization correctness, benchmark forensics. If your cluster is Grace-Blackwell and the problem is "it's slower or stranger than it should be," it's in scope. Running 4–8 nodes in production? The $7,500 cluster audit is the full-stack version of this — every layer, same guarantee.
Book it.
One email starts it: model, config, node count, symptom, the number you're stuck at.
contact@anantus.net →Response within one business day. If we're at capacity you get a start date, not silence.