pragma.vision Technology observatory & strategic foresight

Verification register AI & Agents

Readiness verdict

Claude 3 family + Claude 3.5 Sonnet (Anthropic)

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

As of
2026-10-05
Revision
3
Method
v1.4.0

Current reading

Readiness band and full integer triple

AI-assisted assembly · derived results

Readiness band

Watch

Primary summary from verified readiness

Confidence
93% · stale
Computed at
2026-10-05T03:28:04.532857+00:00
Claimed
55

Public ambition and stated capability

Reported
35

Observed practitioner reporting

Verified
35

Independently supported evidence

Gap
+20

Claimed leads verified

Evidence strength Growing

Decision

What the current evidence supports

Agent-assisted assessment · independently reviewed · 2026-10-05

Favor actively supported models for new deployments; retain Opus 3 when its particular behavior is needed and continued access is confirmed.

Why
These models demonstrated useful supervised capabilities, but the scoped Sonnet and Haiku versions have retired from ordinary Anthropic API service. Opus 3 has a paid-chat and API-by-request exception; substantial task failures remain.
Next
Confirm Opus 3 access and continuity terms, test representative tasks with human review, and migrate retired Sonnet and Haiku integrations.

Constraints

Blockers

  • Retired Sonnet and Haiku versions lack ordinary Anthropic API access.
  • Opus 3 API access is by request, with no promise of indefinite service.
  • Task failures, hallucinations and overfitted patches require output checks.
  • Evidence varies by model, task, prompt and agent scaffold.
  • Computer use in the October Sonnet version was an error-prone public beta.
  • Partner-platform lifecycle schedules may differ.

Evidence summary

Derived counts

AI-assisted assembly

Total
11
Tier 1
10
Tier 2
1
Tier 3
0
Supports
6
Contradicts
2
Context
3
Latest observed
2026-04-20

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

Publication record

Revisions

Revision 3

  1. 2026-07-23 Reading moved from unknown to unknown.
  2. 2026-09-22 Reading moved from unknown to unknown.
  3. 2026-10-05 Reading moved from unknown to watch.

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.