models · stepfun-ai

Step-3.7-Flash

image-text-to-text model harvested from Hugging Face (trending). Catalogued 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.

Method	hf-published-sha256

Re-check this yourself → Source → Full record →

What is recorded

Publisher	stepfun-ai
Origin	China
Licence	apache-2.0
Parameters	201,365,316,160 (201.4B)
Context window	262,144 tokens
Modality	text+image+video->text
Native precision	BF16
Pulls recorded here	50,207
Downloads reported upstream	19,350

Reported scores

GPQA Diamond	80.9
IFBench	67.3
Humanity's Last Exam	21.4

Reported by a third party and recorded here, not re-run by us. Source: artificialanalysis.ai/api/v2.

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.

Reported to local-ai-registry

Hardware	Engine	Build measured	Decode, 1 stream	Label	Reported by
Apple M5 Max 128GB	llama.cpp	GGUF IQ3_XXS	60.3 tok/s @ 8,256	candidate	exo-postgres
Apple M5 Max 128GB	llama.cpp	GGUF iq4xs	59.6 tok/s @ 8,256	candidate	exo-postgres
Apple M5 Max 128GB	llama.cpp	GGUF IQ4_XS	59.6 tok/s @ 8,256	candidate	exo-postgres
Apple M3 Ultra 96GB 80-core GPU	llama.cpp	GGUF IQ4_XS	59.0 tok/s @ 8,256	candidate	exo-postgres
Apple M5 Max 128GB	llama.cpp	GGUF IQ3_XXS	58.2 tok/s @ 8,256	candidate	exo-postgres
Apple M5 Max 128GB	llama.cpp	Q4_K_S	58.2 tok/s @ 8,256	candidate	exo-postgres
Apple M3 Ultra 96GB 80-core GPU	llama.cpp	GGUF IQ3_XXS	57.7 tok/s @ 8,256	candidate	exo-postgres
Apple M3 Ultra 96GB 80-core GPU	llama.cpp	Q4_K_S	57.7 tok/s @ 8,256	candidate	exo-postgres
Apple M5 Max 128GB	llama.cpp	GGUF UD-IQ2_XXS	56.5 tok/s @ 8,256	candidate	exo-postgres
Apple M5 Max 128GB	llama.cpp	GGUF Q3_K_M	56.3 tok/s @ 8,256	candidate	exo-postgres

From local-ai-registry (MIT), commit 124959d: 44 runs across 8 machines, 0 validated — model revision and runtime pinned, launch accepted — and 44 candidate, their label for useful evidence without a reproducible-launch promise. 28 more not listed.

Measured by local.ai

Hardware	Build measured	Decode @ 8K	Memory @ 8K	Decode mode
NVIDIA RTX PRO 6000 Blackwell 96GB	IQ3_XXS	158 tok/s	47.5 GB	ordinary
NVIDIA RTX PRO 6000 Blackwell 96GB	UD-IQ2_XXS	144 tok/s	75.6 GB	ordinary
NVIDIA RTX PRO 6000 Blackwell 96GB	UD-IQ2_M	142 tok/s	75.7 GB	ordinary
MacBook Pro M5 Max 128GB · 40-core GPU	IQ3_XXS	66.5 tok/s	48.2 GB	ordinary
MacBook Pro M5 Max 128GB · 40-core GPU	Q3_K_M	62.5 tok/s	106.0 GB	ordinary
MacBook Pro M5 Max 128GB · 40-core GPU	UD-IQ2_XXS	61.9 tok/s	74.6 GB	ordinary
MacBook Pro M5 Max 128GB · 40-core GPU	UD-IQ2_M	60.1 tok/s	74.7 GB	ordinary
MacBook Pro M5 Max 128GB · 40-core GPU	IQ4_XS	59.8 tok/s	117.2 GB	ordinary

Measured by local.ai, not by us: 33 runs, 25 more not listed. Each names its engine pinned by image digest and the exact serve command, in the full record.

Elsewhere in the registry

All models → Browse with filters → What will run on my hardware? →

Registry record as of 2026-09-01. 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.