Current report presentation · catalog updated 22 September 2026
Status · 22 September 2026 (Europe/Istanbul). The 7B class opened on 21 September 2026. Any registered account may scan the repositories on the eligible list of the scope page — exact Apache-2.0 revisions, listed there with their status — within the free allowance of 20 browser scans and five distinct source models per calendar month. 7B-class repositories run as single-model quantization simulations. A comparison of two 7B-class checkpoints is currently accepted by the submission form but fails inside the worker sandbox (under investigation, 22 September 2026); please do not submit one until this sentence is removed. HuggingFaceTB/SmolLM3-3B-Base and HuggingFaceTB/SmolLM3-3B are listed on the register, but the current instrument release does not support their tokenizer contract, so those submissions fail as well. Scans run on Tetracta-operated local GPU workers; no cloud GPU is used in this beta. Models above the 7B class are not scanned in this beta. External customer acceptance remains pending; results are research-stage and payments are disabled.
The examples below are genuine measurements rendered through the current VG1 customer-report code. Reports 07–13 were produced by the live service on 21–22 September 2026 and each links to its own operational receipt; report 14 is a labelled research run without a receipt; the earlier examples are unsigned public copies of retained measurements. Original measurement times and evidence limits remain in each report; the catalog date is not a new measurement date. Knowledge scans remain unavailable. No model-quality ranking, safety verdict, causal localization or risk grade is supplied. Hallucination, undertraining and overtraining remain research questions, not measured product features.
Report contract rs-1.7 / mv-1.4; presentation rp-1.3. Catalog updated 22 September 2026.
Seven new scans of five published checkpoints, with repeat and training-seed controls.
Both same-checkpoint repeat comparisons showed no difference. All four within-family seed comparisons and both cross-family comparisons showed response and output changes. The cross-family result therefore does not isolate an architecture effect or rank model quality.
Open the results table and ten reports → · Pinned report files
SmolLM2-360M → SmolLM2-360M-Instruct
Live-service report, 22 September 2026. Difference observed at 29 of 29 evaluated positions; text output withheld (output contracts differ); overall relative weight change 7.938%. Combined artifact change.
Qwen2.5-1.5B → Qwen2.5-Coder-1.5B-Instruct
Live-service report, 22 September 2026. Difference observed at 25 of 25 evaluated positions; text output withheld; overall relative weight change 62.96%. Combined artifact change.
Qwen2.5-1.5B → Qwen2.5-Math-1.5B-Instruct
Live-service report, 22 September 2026. Difference observed at 25 of 25 evaluated positions; text output withheld; overall relative weight change 100.4%. Combined artifact change.
Qwen2.5-7B-Instruct → partial in-memory INT8 weight simulation
Live-service report, 21 September 2026. Difference observed at 25 of 25 evaluated positions; text output changed on the evaluated inputs; overall relative weight change 0.7875%. Partial in-memory simulation, not a deployed quantized artifact.
Qwen2.5-7B → partial in-memory INT8 weight simulation
Live-service report, 22 September 2026. Difference observed at 25 of 25 evaluated positions; text output changed on the evaluated inputs; overall relative weight change 0.7878%. Partial in-memory simulation, not a deployed quantized artifact.
Mistral-7B-Instruct-v0.3 → partial in-memory INT8 weight simulation
Live-service report, 22 September 2026. Difference observed at 29 of 29 evaluated positions; text output changed on the evaluated inputs; overall relative weight change 0.2098% over 291 parameter tensors. Partial in-memory simulation, not a deployed quantized artifact.
Qwen3-4B-Base → Qwen3-4B
Live-service report, 22 September 2026. Difference observed at 33 of 33 evaluated positions; text output withheld (output contracts differ); overall relative weight change 9.491% over 398 parameter tensors. Combined artifact change.
Mistral-7B-v0.3 → Mistral-7B-Instruct-v0.3
Research run, 22 September 2026, same instrument release outside the customer path (the customer path failed for a pair of two 7B checkpoints that day; under investigation). Difference observed; this unsigned copy is rendered by the older standalone template and carries no depth view, so it prints no "N of N" summary line. Its per-block record shows a difference at all 29 evaluated positions, which is how that figure is quoted elsewhere; its own printed "24 probes evaluated" is the size of the evaluated set at its reporting depth, not a count of what differed. Text output withheld (tokenizer contracts differ); overall relative weight change 0.4330% over 291 parameter tensors. Combined artifact change. No operational receipt.
SmolLM2-135M → SmolLM2-135M-Instruct
Recorded comparison; measurement time is unavailable in this record.
Qwen2.5-7B
In-memory 8-bit simulation. Same original measurement as the compact view.
Qwen2.5-7B
The same measurement as the depth view, presented without the depth map; not another experiment.
Qwen2.5-7B → itself
Original same-artifact observations. Repeat and negative-control summaries use the same measurements.
Qwen2.5-7B → itself
Weight-summary output from a separate retained same-artifact qualification; the report is labelled a validation-pending descriptive result. It does not establish changed-model sensitivity.
Llama-3.3-70B-Instruct
A retained 70B in-memory simulation; not a free-beta eligibility grant or a new customer result. No 7B control is transferred to this example.
The original historical content below is retained for traceability. It is not the default current-example catalog. Availability statements are superseded by the current status above.
VG1 has been deployed. Customer scan-to-report acceptance is still pending.
The free first-beta scope is eligible public models up to and including 7B, within the account allowance, for checkpoint comparisons and in-memory quantization simulations. Customer jobs run on RunPod. (Superseded on 22 September 2026: scans run on Tetracta-operated local GPU workers; no cloud GPU is used in this beta.) The service selector defines the supported revisions and scan types. Later Pro access above 7B requires an explicit grant and remains limited to the models and allowance enabled for that account.
The thirteen existing sample reports retain their own dates, versions and recorded evidence. They are not results of the current customer acceptance test. The 7 September sample documentation below is retained as historical context and is superseded in more than its coverage wording. Read it only as a record of what we published on that date. In particular: on 12 September 2026 we corrected the characterization of the unmeasured positions — describing them as lying below an instrument floor was wrong; it was a scanner-coverage limit, and the current report format says "not evaluated" instead. The depth and location wording below, including onset depth and the comparison of a fine-tune with a structural lesion, is not how the current release describes itself, and no claim below is reinstated by its presence here. The standing record is the dated correction page: https://www.tetracta.ai/model-xray/correction/
A report describes its recorded artifacts and measurement conditions; it is not a model-quality ranking, safety certificate or deployment verdict. A quantization simulation does not create a deployable quantized model. Knowledge scans remain unavailable and earlier withdrawn claims remain withdrawn. Hallucination, undertraining and overtraining remain research questions, not measured product features.
Service status · Current scope and report guide · 12 September correction
When a public model was fine-tuned by us on four consecutive decoder blocks at a known depth (LoRA r=8 on blocks LO..LO+3, merged; three depths), the reported onset matched the first fine-tuned block in 3 of 3 cells (blocks are numbered from 1). This does not establish behaviour on fine-tunes produced by others, whose depth and magnitude are unknown.
Frozen B7 preregistration SHA-256: 1028eedb853fb662aeb85a6d5f0404df1dcf99d367088bc2c94a9000a3243977.
Two controls were run first: one artifact, re-serialized into a different shard layout with identical tensor values, was reported as no difference; a known lesion reported a difference at the expected depth. In 3 controlled micro-fine-tune pairs (single model family), every difference this instrument reported was also visible in the model's output text. We have not yet produced a case where the instrument reports a difference that a plain output comparison would miss. Until we do, we make no claim that it can.
Frozen B8-SD preregistration SHA-256: d2dae32fa361f16a9f660241181b4a5bd19ca44338d284b15dbfc57e0c7a6d9d.
Onset localization works on our controlled fine-tunes; added value over plain output comparison is not yet shown.
The reading tells you where a change first becomes observable. It does not tell you how large the change is, or what kind of change it was: in our controlled tests, a fine-tune and a structural lesion at the same location produced the same reading.
Configuration comparison is limited to the compared fields. Numeric-precision fields are not compared here; any file difference is recorded in the artifact identity.
Reports now show categorical onset depth, a depth map and the extent of the observable difference. For all ten reports the difference is already present at the first evaluable position, so it may begin earlier, below the instrument floor; no starting depth is reported for these pairs.
Calibration: exact onset predicted and observed in 6 of 6 controlled cells on each of three public models (Qwen2.5-0.5B-Instruct, SmolLM2-1.7B, TinyLlama-1.1B-Chat-v1.0); cross-board agreement on onset 4 of 4 cells. These controlled changes do not establish performance on changes that arise naturally. The first four profile positions, the embedding output and the outputs of blocks 1 to 3, are below the instrument floor.
16 of 16 A = B comparisons completed across four models and two card classes: 0 observed differences and 0 output-text changes; two-sided exact 95% Clopper–Pearson interval 0–20.59%. This describes the tested matrix, not a population error rate.
| # | Pair · revisions | Card · runtime | Pair relation | Difference observed | Output text changed | Onset | Probes evaluated | Report |
|---|---|---|---|---|---|---|---|---|
| 1 | Qwen/Qwen2.5-0.5B@ 060db6499f32 → Qwen/Qwen2.5-0.5B-Instruct@ 7ae557604adf | RTX 5070 TiPyTorch 2.11.0+cu128 / Transformers 5.10.1 | combined artifact change (configs differ) | yes | withheld | already present at the first evaluable position | 24 | Open report → |
| 2 | Qwen/Qwen2.5-0.5B@ 060db6499f32 → Qwen/Qwen2.5-0.5B-Instruct@ 7ae557604adf | RTX 4060PyTorch 2.9.1+cu128 / Transformers 5.5.4 | combined artifact change (configs differ) | yes | withheld | already present at the first evaluable position | 24 | Open report → |
| 3 | Qwen/Qwen2.5-1.5B@ 8faed761d45a → Qwen/Qwen2.5-1.5B-Instruct@ 989aa7980e4c | RTX 5070 TiPyTorch 2.11.0+cu128 / Transformers 5.10.1 | combined artifact change (configs differ) | yes | withheld | already present at the first evaluable position | 24 | Open report → |
| 4 | Qwen/Qwen2.5-1.5B@ 8faed761d45a → Qwen/Qwen2.5-1.5B-Instruct@ 989aa7980e4c | RTX 4060PyTorch 2.9.1+cu128 / Transformers 5.5.4 | combined artifact change (configs differ) | yes | withheld | already present at the first evaluable position | 24 | Open report → |
| 5 | HuggingFaceTB/SmolLM2-1.7B@ effd688a1292 → HuggingFaceTB/SmolLM2-1.7B-Instruct@ 31b70e2e869a | RTX 5070 TiPyTorch 2.11.0+cu128 / Transformers 5.10.1 | combined artifact change (configs differ) | yes | withheld | already present at the first evaluable position | 24 | Open report → |
| 6 | HuggingFaceTB/SmolLM2-1.7B@ effd688a1292 → HuggingFaceTB/SmolLM2-1.7B-Instruct@ 31b70e2e869a | RTX 4060PyTorch 2.9.1+cu128 / Transformers 5.5.4 | combined artifact change (configs differ) | yes | withheld | already present at the first evaluable position | 24 | Open report → |
| 7 | TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T@ 59f6f375b26b → TinyLlama/TinyLlama-1.1B-Chat-v1.0@ fe8a4ea1ffed | RTX 5070 TiPyTorch 2.11.0+cu128 / Transformers 5.10.1 | combined artifact change (configuration identical in the compared fields; tokenizer semantics differ) | yes | withheld | already present at the first evaluable position | 24 | Open report → |
| 8 | TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T@ 59f6f375b26b → TinyLlama/TinyLlama-1.1B-Chat-v1.0@ fe8a4ea1ffed | RTX 4060PyTorch 2.9.1+cu128 / Transformers 5.5.4 | combined artifact change (configuration identical in the compared fields; tokenizer semantics differ) | yes | withheld | already present at the first evaluable position | 24 | Open report → |
| 9 | Qwen/Qwen2.5-0.5B-Instruct@ 7ae557604adf → KingNish/Qwen2.5-0.5b-Test-ft@ 7917c64eeab3 | RTX 5070 TiPyTorch 2.11.0+cu128 / Transformers 5.10.1 | combined artifact change (no difference recorded in the compared configuration fields) | yes | yes | already present at the first evaluable position | 24 | Open report → |
| 10 | HuggingFaceTB/SmolLM2-1.7B-Instruct@ 1dfd32118ef1 → motexture/SmolLCoder-1.7B-Instruct@ d9b7eb7b380d | RTX 5070 TiPyTorch 2.11.0+cu128 / Transformers 5.10.1 | combined artifact change (no difference recorded in the compared configuration fields) | yes | yes | already present at the first evaluable position | 24 | Open report → |
| 11 | Qwen/Qwen2.5-0.5B-Instruct@ 7ae557604adf → tetracta/qwen2.5-0.5b-instruct-lora-blocks-11-14local checkpoint SHA-256 3bef74e68554 | RTX 5070 TiPyTorch 2.11.0+cu128 / Transformers 5.10.1 | combined artifact change (configuration identical in the compared fields) | yes | yes | first observable at decoder block 12 of 24 | 24 | Open report → |
| 12 | Qwen/Qwen2.5-0.5B-Instruct@ 7ae557604adf → tetracta/qwen2.5-0.5b-instruct-lora-blocks-5-8local checkpoint SHA-256 b3100da75e1f | RTX 5070 TiPyTorch 2.11.0+cu128 / Transformers 5.10.1 | combined artifact change (configuration identical in the compared fields) | yes | yes | first observable at decoder block 6 of 24 | 24 | Open report → |
| 13 | Qwen/Qwen2.5-0.5B-Instruct@ 7ae557604adf → tetracta/qwen2.5-0.5b-instruct-lora-blocks-18-21local checkpoint SHA-256 8cc3c44c6b45 | RTX 5070 TiPyTorch 2.11.0+cu128 / Transformers 5.10.1 | combined artifact change (configuration identical in the compared fields) | yes | yes | first observable at decoder block 19 of 24 | 24 | Open report → |
Rows 9–10: numeric precision (torch_dtype) differs; recorded in the artifact identity.
Rows 11–13 are controlled fine-tunes produced for calibration; not for use. Their B identifiers are local checkpoint SHA-256 values, not uploaded model revisions. No checkpoint files are included.
Rows 1–8 have different output contracts, so output text is withheld. Rows 9–10 share a comparable output contract: the recorded text was compared and differs. The depth maps record a difference at every evaluable position. No weights-only or pure fine-tuning attribution is made. Training claims for rows 9–10 are the authors’ descriptions and were not independently checked.
Rows 9–10 were also run on RTX 4060 with the same difference, output-text status and onset status; their card class and runtime are recorded in the full-revision index. The sample-pair agreement of 6 of 6 pairs is separate from the controlled experiment’s 4 of 4 cells.
TinyLlama configuration comparison is identical in the compared fields; tokenizer semantics differ. Commercial service admission rights review is still pending, separately from this publication. Row 10 uses the SmolLM2-1.7B-Instruct revision of 4 December 2024 to match the derivative’s tokenizer length and chat contract.
Read Scope & Limitations and the correction record. Numerical profiles remain private. No numerical equivalence across card classes or runtimes, quality verdict, or fitness-for-use assessment is claimed.