Qwen2.5 VL 72B Instruct
Qwen · released Jan 27, 2025 · Qwen/Qwen2.5-VL-72B-Instruct
- Type
- Open weightsCustom licence
- Params
- 73.4B
- Context
- 128K
about 96K words of context · download allowed, licence restricts use
Our take
Written Aug 3, 2026Qwen2.5 VL is a large vision-language model that accepts text and images and can handle up to 128,000 tokens in a single request. Its weights can be downloaded under a custom restricted licence, and it is the widest-context variant in its family at this scale.
Choose this for vision-language tasks with long documents and images, where the 128,000-token request limit matters. It suits cost-sensitive inference if you can access the cheapest host, and scenarios where downloadable weights matter more than a fully permissive licence. Skip it if you need verified quality benchmarks, unrestricted commercial redistribution, or flat input-output pricing.
The case for it
- Widest context window among Qwen2.5 VL variants at this scale, at 128,000 tokens.
- Cheapest input among current tracked offers, well under half the next option.
- Throughput measured at 27 tokens per second on one host and 26 on another.
The case against it
- No verified quality benchmarks in our data — no Elo, MMLU or other scores listed.
- Custom restricted licence, not Apache 2.0 or MIT; commercial flexibility is limited compared with fully permissive alternatives.
- Output cost on the cheapest host is three times the input cost, a steeper multiplier than flat-rate alternatives; throughput unverified for half the tracked offers.
How good is it?
We hold no score for this model.
We look for every model we track on every board we watch, and none of them has turned up Qwen2.5 VL 72B Instruct — so there is no intelligence, coding, agentic or writing score to show you, not a low one, none.
Can you run it yourself?
- Fits in memory
- weights load entirely on the card
- Spills to system RAM
- some weights offload; much slower
- Too large
- will not load even with offload
- est
- size is calculated; the verdict could change by 10%
GeForce RTX 4090 · 24 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Apple M1 Pro (16-core GPU) · 32 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Comfortable fit
Apple M2 Max (38-core GPU) · 96 GB
Room to spare. 22.5 GB spare means a 10% error in the size would not change the answer.
Memory use by level
Against a 24 GB card.
All 3 sizes here are calculated, not measured. We hold no measured file for this model, so each size comes from the parameter count and every verdict above inherits that uncertainty.
Check against your own machine → · All 71 devices, with every size →
Or rent it from someone else
Cheapest of 4 live listings. Picked at the widest standard context we hold, within one quantisation slice, so the numbers beside it are a price one host actually charges.
- per 1M tokens
- $0.80 in / $1.00 out
- Context served
- 128K
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| Nebius AI Studiofp8 | $0.25 / $0.75 | 32K | 25 tok/s | No | No | Confirmed |
| Novita AI | $0.80 / $0.80 | 33K | not measured | Unknown | Unknown | Unknown |
| OpenRouter | $0.80 / $1.00 | 128K | not measured | Unknown | Unknown | Unknown |
| Parasailfp8 | $0.80 / $1.00 | 128K | 30 tok/s | No | No | Confirmed |
Nebius AI Studio
Novita AIAcross the 4 listings we hold: 2 say they do not train on prompts, 0 say they do and 2 do not say. 2 appear in the zero-retention registry we check; the rest are unknown to us rather than confirmed either way.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
- ✓
- Supported
- ✗
- Not supported
- Not published
- host gave no parameter list
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| Nebius AI Studiofp8 | ✗ | ✓ | ✓ |
| Novita AI | |||
| OpenRouter | ✗ | ✓ | ✓ |
| Parasailfp8 | ✗ | ✓ | ✓ |
Tool calling: 0 of 4 listings say yes, 3 say no, 1 publishes no parameter list. JSON output: 3 of 4 listings say yes, 1 publishes no parameter list. Strict schema: 3 of 4 listings say yes, 1 publishes no parameter list.
Models people weigh against Qwen2.5 VL 72B Instruct
When we formed this view
Dates behind this page
Prices last checked 38h ago
What we do not know about this model yet
- We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
- No board we watch has turned up a score, so we hold no quality figures at all.
- 1 of 4 listings publish no parameter list, so what their API accepts is unknown to us.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 2 of 4 listings do not say whether they train on prompts.
Licence and identifiers
What the licence allowsCustom licence, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.
Licence
Custom licence
This model ships custom license terms that don't map to a known template. We haven't parsed them, so commercial use, redistribution and derivatives are unverified — review the original terms before shipping.
Identifiers
- Hugging Face
- Qwen/Qwen2.5-VL-72B-Instruct
- Architecture
- Dense
- Modality record
- text+image->text
- Catalogue slug
- qwen-qwen2-5-vl-72b-instruct