models · internscience

Agents-A1

Catalogued from local.ai local-inference benchmark index at Quarantine — unverified, pending attestation.

Compare Verify

What this registry has checked

Quarantine. We hold the publisher’s own published hash for these weights and have recorded it against this entry. We have not downloaded and hashed the weights ourselves, so the hash is the publisher’s claim, not our measurement.
Methodhf-published-sha256

What is recorded

Publisherinternscience
OriginNot recorded. The harvest wrote its own source name into this field, which tells us nothing about jurisdiction.
LicenceApache-2.0
Parameters35,107,181,936 (35.1B)
Native precisionBF16
Downloads reported upstream9,963

Measured on real hardware

None of these runs are ours. Each figure is the reporter’s, linked to their own evidence, under their own trust label. The number shown is single-stream decode — one user, concurrency 1 — because aggregate throughput across many users is a different quantity and reads as far faster than anyone will see.

Measured by local.ai

HardwareBuild measuredDecode @ 8KMemory @ 8KDecode mode
NVIDIA RTX PRO 6000 Blackwell 96GBq4km267 tok/s27.4 GBordinary
NVIDIA RTX PRO 6000 Blackwell 96GBFP8197 tok/s90.7 GBordinary
NVIDIA RTX 6000 Ada 48GBq4km183 tok/s27.2 GBordinary
NVIDIA RTX 6000 Ada 48GBFP8132 tok/s45.0 GBordinary
MacBook Pro M5 Max 128GB · 40-core GPUq4km114 tok/s27.4 GBordinary
MacBook Pro M4 Max 128GB · 40-core GPUq4km90.8 tok/s27.4 GBordinary
Mac Studio M3 Ultra 96GB · 80-core GPUq4km89.8 tok/s27.4 GBordinary
MacBook Pro M5 Max 128GB · 40-core GPU8bit86.1 tok/s39.3 GBordinary

Measured by local.ai, not by us: 19 runs, 11 more not listed. Each names its engine pinned by image digest and the exact serve command, in the full record.

Elsewhere in the registry

Registry record as of 2026-09-26. Identity, licence and parameter count are properties of a release and do not move; tier can, and the verify link above re-checks it against the live registry rather than this page.