Verification register AI & Agents
Current reading
Readiness band and full integer triple
AI-assisted assembly · derived results
Readiness band
Ready
Primary summary from verified readiness
- Confidence
- 69% · fresh
- Computed at
- 2026-07-21T07:36:39.532946+00:00
- Claimed
- 72
- Reported
- 65
- Verified
- 56
- Gap
- +16
Public ambition and stated capability
Observed practitioner reporting
Independently supported evidence
Claimed leads verified
Evidence strength Strong
Decision
What the current evidence supports
Human editorial judgment · 2026-07-21
Track; not yet
- Why
- Genuinely differentiated on cost and token efficiency (4.2x fewer output tokens than Opus 4.8 on agentic tasks, $2/$6 per-million pricing vs $5/$25 for Opus 4.8) and notable for training partly on real Cursor developer session data, but it trails the top three coding models on raw SWE-bench Pro capability, is not yet available in the EU, its fastest inference path is not yet deployed, and its most publicized capability claim (a math breakthrough) is unverified — too many open threads for adoption today.
- Next
- Re-evaluate once EU access opens, xAI's native GB300 inference stack ships (claimed to double+ current throughput), and the viral hypercontractivity math claim either gets a peer-reviewed writeup or is dropped/retracted; re-check independent benchmark position (Artificial Analysis Intelligence Index) against Opus/Fable/GPT lineage at that point.
Constraints
Blockers
No named blocker is present in the current public projection.
Evidence summary
Derived counts
AI-assisted assembly
- Total
- 5
- Tier 1
- 0
- Tier 2
- 5
- Tier 3
- 0
- Supports
- 2
- Contradicts
- 1
- Context
- 2
- Latest observed
- 2026-07-10
Counts and dates only. Raw signals, private excerpts, trust records, and internal corpus material are not published here.
Publication record
Revisions
Initial public reading
- 2026-07-21 Reading moved from ready to ready.