pragma.vision Your technology observatory

Verification register AI & Agents

Readiness verdict

NVIDIA Nemotron 3 Nano

A dated reading of what is claimed, reported, and independently verified in the current evidence.

As of
2026-06-28
Revision
1
Method
v1.0.0

Current reading

The readiness gap, in one scan

AI-assisted assembly · derived results

Claimed
85

Public ambition and stated capability

Reported
79

Observed practitioner reporting

Verified
69

Independently supported evidence

Gap
+16

Claimed minus verified

Evidence strength Critical

Decision

What the current evidence supports

Human editorial judgment · 2026-06-28

Proceed with caution

Why
Strong open agentic-reasoning accuracy and throughput make it compelling for server/edge-GPU agent workloads, but the 31.6B footprint, VRAM-gated long context, non-standard license, and mid SWE-Bench mean it is not a drop-in mobile on-device model and needs hardware/licensing due diligence.
Next
Pilot the BF16 weights via vLLM/TRT-LLM on available H100/B200 or Jetson Thor / DGX Spark edge hardware for agentic-reasoning workloads (or trial hosted at ~$0.05/$0.20 per Mtok via OpenRouter); benchmark throughput at the intended context length and review the Nemotron Open Model License before production.

Constraints

Blockers

No named blocker is present in the current public projection.

Evidence summary

Derived counts

AI-assisted assembly

Total
6
Tier 1
0
Tier 2
1
Tier 3
5
Supports
4
Contradicts
2
Context
0
Latest observed
2026-06-24

Counts and dates only. Raw signals, private excerpts, trust records, and internal corpus material are not published here.

Publication record

Revisions

Initial public reading

This is the initial public reading. No earlier readiness change is recorded.

Your opinion

Tell us anything.

What works, what doesn't, what's missing — especially about our watches, lenses, and the register itself. Anonymous is fine; leave an email if you'd like a reply.