Skip to content
Get an API key

Live · serverless · 2 regions

Qwen2.5 VL 72B Instruct

by Alibaba

A 72B vision-language model that reads charts, screenshots, scanned forms and long documents.

  • vision
  • chat
  • FP8
Serverless priceUSD
Input
$0.60per 1M tokens
Output
$0.80per 1M tokens

Billed per token, no minimums.

Reference

Specifications

Model ID
qwen2-5-vl-72b
Endpoint
POST /v1/chat/completions
Publisher
Alibaba
Tasks
vision · chat
Served context
32K tokens
Quantization
FP8
Decode speed
80 tok/smedian, one request
p50 first token
330 mslast 24 hours
Regions
us-east · us-west
Dedicated
On request

Last 24 hours

Latency and speed

Median of each 15-minute step on the shared pool, measured at the API edge. Time to first token includes queueing; decode speed is for one request.

Time to first token · ms

330 msmedian

p95 357 ms

Decode speed · tok/s

80 tok/smedian

p5 76 tok/s

Show the data
Median of each 15-minute step over the last 24 hours, newest first
StepTime to first tokenDecode speed
15m ago332 ms78 tok/s
30m ago337 ms81 tok/s
45m ago341 ms80 tok/s
1h ago325 ms79 tok/s
1h 15m ago333 ms77 tok/s
1h 30m ago342 ms77 tok/s
1h 45m ago357 ms75 tok/s
2h ago353 ms78 tok/s
2h 15m ago338 ms80 tok/s
2h 30m ago418 ms72 tok/s
2h 45m ago357 ms76 tok/s
3h ago340 ms78 tok/s
3h 15m ago345 ms79 tok/s
3h 30m ago358 ms78 tok/s
3h 45m ago330 ms79 tok/s
4h ago327 ms82 tok/s
4h 15m ago337 ms82 tok/s
4h 30m ago340 ms78 tok/s
4h 45m ago342 ms80 tok/s
5h ago336 ms80 tok/s
5h 15m ago356 ms77 tok/s
5h 30m ago330 ms82 tok/s
5h 45m ago325 ms79 tok/s
6h ago331 ms79 tok/s
6h 15m ago323 ms81 tok/s
6h 30m ago351 ms79 tok/s
6h 45m ago346 ms78 tok/s
7h ago322 ms79 tok/s
7h 15m ago333 ms78 tok/s
7h 30m ago331 ms81 tok/s
7h 45m ago345 ms80 tok/s
8h ago340 ms79 tok/s
8h 15m ago336 ms81 tok/s
8h 30m ago318 ms81 tok/s
8h 45m ago342 ms76 tok/s
9h ago315 ms83 tok/s
9h 15m ago339 ms80 tok/s
9h 30m ago327 ms81 tok/s
9h 45m ago334 ms81 tok/s
10h ago321 ms82 tok/s
10h 15m ago311 ms83 tok/s
10h 30m ago330 ms80 tok/s
10h 45m ago334 ms82 tok/s
11h ago325 ms79 tok/s
11h 15m ago310 ms85 tok/s
11h 30m ago323 ms80 tok/s
11h 45m ago330 ms78 tok/s
12h ago330 ms79 tok/s
12h 15m ago319 ms83 tok/s
12h 30m ago325 ms82 tok/s
12h 45m ago315 ms80 tok/s
13h ago334 ms80 tok/s
13h 15m ago305 ms82 tok/s
13h 30m ago328 ms82 tok/s
13h 45m ago333 ms81 tok/s
14h ago315 ms81 tok/s
14h 15m ago315 ms82 tok/s
14h 30m ago332 ms78 tok/s
14h 45m ago320 ms82 tok/s
15h ago325 ms80 tok/s
15h 15m ago322 ms79 tok/s
15h 30m ago327 ms82 tok/s
15h 45m ago324 ms81 tok/s
16h ago329 ms81 tok/s
16h 15m ago322 ms82 tok/s
16h 30m ago319 ms80 tok/s
16h 45m ago311 ms82 tok/s
17h ago327 ms81 tok/s
17h 15m ago334 ms81 tok/s
17h 30m ago314 ms84 tok/s
17h 45m ago326 ms78 tok/s
18h ago322 ms81 tok/s
18h 15m ago317 ms82 tok/s
18h 30m ago312 ms80 tok/s
18h 45m ago312 ms81 tok/s
19h ago318 ms84 tok/s
19h 15m ago455 ms65 tok/s
19h 30m ago308 ms84 tok/s
19h 45m ago338 ms77 tok/s
20h ago337 ms78 tok/s
20h 15m ago311 ms83 tok/s
20h 30m ago336 ms81 tok/s
20h 45m ago323 ms80 tok/s
21h ago331 ms79 tok/s
21h 15m ago325 ms83 tok/s
21h 30m ago336 ms79 tok/s
21h 45m ago318 ms82 tok/s
22h ago339 ms81 tok/s
22h 15m ago350 ms79 tok/s
22h 30m ago321 ms80 tok/s
22h 45m ago320 ms82 tok/s
23h ago334 ms80 tok/s
23h 15m ago326 ms81 tok/s
23h 30m ago325 ms80 tok/s
23h 45m ago345 ms77 tok/s
24h ago349 ms80 tok/s

API

Make a request

The same request shape as every other model on /v1/chat/completions. Switching between them is a change to one string.

curl https://api.weir.example/v1/chat/completions \
  -H "Authorization: Bearer $WEIR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen2-5-vl-72b",
    "messages": [
      { "role": "user", "content": "What does a KV cache do?" }
    ]
  }'
POST /v1/chat/completionsapi.weir.example

Dedicated

Need reserved capacity?

Qwen2.5 VL 72B Instruct runs on the shared serverless pool. Dedicated replicas for it are set up on request, with the same API and the same model ID.

Talk to an engineer

Compare

Similar models