Qwen2.5-VL-72B-Instruct qwen2.5-vl-72b-instruct

qwen · qwen vision

unranked — no quality signals yet

params
73.4B
arch
dense
context
125k
license
other open weights

# architecture

attn shared GQA 64q/8kv feed-forward (dense) every parameter, every token out ×80 layers ctx 128,000 73.4B — no routing
layers
80
d_model
8192
heads
64q / 8kv
head dim
128
vocab
152064
family
qwen2_5_vl

signal path of one layer, generated from the registry's structured fields — dimension lines quote the model's real numbers; an MoE trace forks at the router, a dense trace runs straight through. Geometry fields come from the repo's config.json.

# trend

OGM score · last 105 days
OGM score over 105 days

methodology changed during this history window; score movement across that boundary is not model movement. See methodology v6.

# benchmarks

benchmarkscoresourcedate
Ai2d 0.9 / 1 LLM Stats —
Aitz-em 0.8 / 1 LLM Stats —
Android-control-high-em 0.7 / 1 LLM Stats —
Android-control-low-em 0.9 / 1 LLM Stats —
Androidworld-sr 0.3 / 1 LLM Stats —
Cc-ocr 0.8 / 1 LLM Stats —
Chartqa 0.9 / 1 LLM Stats —

# where to run

providerquantctx$/M in$/M out$/M cacheprice srctpsuptime✓
Nebius 32k $0.25$0.75— via litellm — —
Novita 32k $0.80$0.80— via litellm — —
Parasailfp8 125k $0.80$1.00$0.40 via openrouter — 100.0%
OVHcloud 32k $1.01$1.01— ovhcloud 31 — ✓
Together 32k $1.95$8.00— via litellm — —

sorted by blended price ((3·input + output) / 4 per 1M) · ✓ = the provider's own catalog confirms the offer · "via …" prices are what the aggregator routing the offer charges, not the provider's own list price

source aliases
llmstats
qwen2.5-vl-72b
openrouter
qwen/qwen2.5-vl-72b-instruct