Tech Review 001 · v1

GPT-5.6 vs Claude Fable 5

One versioned collection connecting seven frozen AI game-dev tests to 28 run ledgers, public integrity evidence, and a real protected Builder research package.

BUILDER PACKAGE · $29.99/MONTH

Audit the workflow behind Tech Review 001.

7 exact prompts · 28 available raw ledgers · 4 configurations · cost/time exports

Get the Builder package Inspect free proof

7frozen prompts
28run ledgers
4configurations
publishededitorial state

Public campaign capture

Public proof

Inspect the claim before buying the package.

The free snapshot exposes model/configuration identity, status, cost and pricing provenance, wall-clock time, processed tokens, artifact and source-record hashes, caveats, and missing evidence. It never exposes exact prompt text or private source files.

Open public comparison →Read methodology Browse 28 evidence records Open immutable manifest

Builder package inventory

Only what is present. Nothing implied.

The ZIP is assembled from the pinned harness commit and delivered only through an authenticated, entitlement-checked endpoint.

Prompts7

Seven exact, frozen prompt files used by this collection.

Run data28

Complete available raw harness JSON ledgers for the selected runs.

Configurations4

Fable Medium, Sol Ultra, Terra Ultra, and Luna extra-high records.

Exports2 CSV

Aggregate plus per-run cost, token, and workflow-time data.

Evidence controlsMatrix

Missing-evidence matrix, public/private manifests, and checksums.

Version pinv1

3f14507a8ce536d68a1f472f34f4c1a64c511f82

Explicitly not included

Known limits stay visible.

Private source artifacts, source code, and Blender files
Complete community preference data or an inferred winner
Repeat-run reliability evidence
Failures or logs that were not present in the reviewed ledgers

Protected Builder delivery

Unlock the exact prompt and ledger package.

Builder includes this exact inventory; the verified community Discord remains free with every account.

Get 7 prompts + 28 ledgers