Verification register Compute & Web Infra
Current reading
Readiness band and full integer triple
AI-assisted assembly · derived results
Readiness band
Ready
Primary summary from verified readiness
- Confidence
- 95% · fresh
- Computed at
- 2026-10-01T01:16:59.403275+00:00
- Claimed
- 75
- Reported
- 65
- Verified
- 55
- Gap
- +20
Public ambition and stated capability
Observed practitioner reporting
Independently supported evidence
Claimed leads verified
Evidence strength Strong
Decision
What the current evidence supports
Agent-assisted assessment · independently reviewed · 2026-10-01
Consider controlled production adoption for tested fixed-weight inference configurations.
- Why
- Production routing gains and inspectable benchmarks are credible, but v1.5.0 has request failures, missing prompt logprobs and unreliable RL weight-version reporting.
- Next
- Pin the release and backend; avoid SGLang sidecar and DCP paths, use aggregated serving when prompt logprobs are required, and defer RL weight updates until correctness is demonstrated.
Constraints
Blockers
- v1.5.0 SGLang sidecar inference fails on every request.
- SGLang DCP workers crash when --dcp-size is enabled.
- Disaggregated SGLang omits requested prompt logprobs.
- vLLM RL weight-version reporting can misidentify stale weights as current.
- Backend and dependency pins complicate upgrades.
- Performance depends on model, traffic reuse and GPU interconnect.
- Current failure recovery and optional features require configuration-specific testing.
Evidence summary
Derived counts
AI-assisted assembly
- Total
- 11
- Tier 1
- 9
- Tier 2
- 2
- Tier 3
- 0
- Supports
- 7
- Contradicts
- 1
- Context
- 3
- Latest observed
- 2026-09-18
Counts and dates only. Raw signals, private excerpts, trust records, and internal corpus material are not published here.
Publication record
Revisions
Revision 3
- 2026-07-19 Reading moved from mature to mature.
- 2026-09-22 Reading moved from mature to mature.
- 2026-10-01 Reading moved from mature to ready.