Bind the source.
Qwen3.8-27B is the source. Mercury is a derived target, not a renamed public model.
Model track
Mercury
An iPhone-fit fine-tuned model derived from Qwen3.8-27B. In training. Device, quality, and release validation remain pending.
Mercury
Fit is a claim only after validation.
Mercury is being prepared for named iPhone configurations. The exact device envelope, model quality, and release status remain subject to validation.
Benchmark
No number without its conditions.
Public benchmark data for Mercury is N/A. When results are available, every value will carry the exact model, suite, runtime, date, and uncertainty needed to interpret it.
Evaluation register / MercuryRelease field · values publish only with complete conditions
Public stateN/A
01 / ENVDevice envelopeNamed iPhone configurations under the intended host
02 / QNTQuantization qualityQuality after the intended compression path
03 / FITOn-device runtimeHost, format, and memory envelope
04 / QLTTask qualityComparison against the stated behavior goal
05 / RELRelease evidencePublic result with complete conditions
01ModelExact revision
02SuiteNamed workload
03RuntimeHost and format
04DateObserved window
05ResultObserved or N/A
Training targets
One track. No public result yet.
Mercury is an iPhone-fit fine-tuned model derived from Qwen3.8-27B. It is in training; device, quality, and release validation remain pending. Not public.
Name the envelope.
Named iPhone configurations are the intended host. Exact fit remains evidence-gated.
Leave the field blank.
Device and quality validation are pending. Public results stay N/A until those conditions exist.