7B class open · free beta
22 September 2026 · Europe/Istanbul
Current status. The 7B class opened on 21 September 2026. The eligible public repositories listed on the scope page (twenty-one exact Apache-2.0 revisions as of 22 September 2026, up to the 7B class) can be scanned by any registered account within the free allowance: 20 browser scans and five distinct source models per calendar month. 7B-class checkpoints run as single-model quantization simulations; a pair of two 7B-class checkpoints is currently accepted by the form but fails in the worker (under investigation, 22 September) — please do not submit one until this note is removed. Scans run on Tetracta-operated local GPU workers; no cloud GPU is used in this beta. A scan reads safetensors weights under their canonical file names; a repository that publishes only a pickle checkpoint or another format (for example GGUF) is declined before a job is created and does not use the allowance. Models above the 7B class are not scanned in this beta. External customer acceptance remains pending; results are research-stage and payments are disabled.
VG1 deployed · external customer acceptance pending (15 September 2026 — superseded by the 22 September entry above)
15 September 2026 · Europe/Istanbul
VG1 is deployed. Internal end-to-end operational acceptance completed on 15 September 2026; external customer acceptance is still pending. The free first-beta scope covers the eligible public models up to and including 7B, within the account allowance. Customer jobs run on RunPod. (Superseded on 22 September 2026: jobs now run on Tetracta-operated local GPU workers; no cloud GPU is used in this beta.) The eligible models are published in the scope statement; exact revisions are pinned when a scan is submitted.
Compare checkpoints.
Read the change.
Inspect a before/after comparison or an in-memory quantization simulation before choosing your next task evaluation. Your report keeps the recorded observation, artifact revisions and comparison limits together.
Before / after comparison
Compare two supported checkpoints under one measurement contract. Read the internal-response finding alongside the output-text status and recorded configuration, tokenizer and generation differences.
Quantization simulation
Compare a source model with its simulated lower-precision state. This is an in-memory experiment: it does not create a quantized model to deploy or measure a packed inference implementation.
A report you can keep
The versioned HTML report includes the finding, context, available repeat and control summaries, and recommended next checks. Where a Depth view is included, categorical observations distinguish unmeasured positions from positions with no observed difference.
A Verify view preserves the finding and output status without the depth observations. Missing verification evidence is labelled, not assumed.
Use it within its scope
A recorded difference is not a quality, risk or capability score. Use the report to guide your own task evaluations. It does not identify edited weights, establish a cause, or certify a deployment.
Access and model coverage
Models above 7B are outside this free first-beta scope. Later Pro access requires an explicit grant and remains limited to the models and allowance enabled for that account; registration alone does not grant access. (22 September 2026: models above the 7B class are not scanned in this beta; any later access to larger models will be announced with a dated update here.)
Private model uploads are not available in this beta. Knowledge scans remain unavailable. Hallucination, undertraining and overtraining are research questions, not measurements offered by this product.
The correction record stays public
The earlier withdrawn findings remain withdrawn. The dated September follow-up explains the coverage and calculation corrections and the limits of the retained Knowledge evidence. Deploying VG1 does not reinstate those claims.
Sample report gallery · Dated corrections
Read each sample's own date, version and evidence status. Historical samples are not relabelled as results from the current release.