Skip to content
Get an API key

Live · serverless · 2 regions

Qwen3 235B A22B

by Alibaba

A large mixture-of-experts model with a switchable thinking mode, strong on multilingual and math-heavy prompts.

  • chat
  • reasoning
  • FP8
Serverless priceUSD
Input
$0.22per 1M tokens
Output
$0.88per 1M tokens

Billed per token, no minimums.

Reference

Specifications

Model ID
qwen3-235b-a22b
Endpoint
POST /v1/chat/completions
Publisher
Alibaba
Tasks
chat · reasoning
Served context
128K tokens
Quantization
FP8
Decode speed
85 tok/smedian, one request
p50 first token
240 mslast 24 hours
Regions
us-east · eu-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

240 msmedian

p95 256 ms

Decode speed · tok/s

85 tok/smedian

p5 82 tok/s

Show the data
Median of each 15-minute step over the last 24 hours, newest first
StepTime to first tokenDecode speed
15m ago239 ms84 tok/s
30m ago248 ms83 tok/s
45m ago251 ms84 tok/s
1h ago247 ms83 tok/s
1h 15m ago260 ms80 tok/s
1h 30m ago246 ms82 tok/s
1h 45m ago255 ms81 tok/s
2h ago255 ms84 tok/s
2h 15m ago247 ms85 tok/s
2h 30m ago244 ms84 tok/s
2h 45m ago240 ms86 tok/s
3h ago259 ms83 tok/s
3h 15m ago245 ms83 tok/s
3h 30m ago249 ms86 tok/s
3h 45m ago256 ms83 tok/s
4h ago242 ms83 tok/s
4h 15m ago239 ms86 tok/s
4h 30m ago246 ms85 tok/s
4h 45m ago259 ms83 tok/s
5h ago255 ms84 tok/s
5h 15m ago255 ms84 tok/s
5h 30m ago245 ms84 tok/s
5h 45m ago242 ms84 tok/s
6h ago252 ms83 tok/s
6h 15m ago254 ms84 tok/s
6h 30m ago245 ms83 tok/s
6h 45m ago249 ms86 tok/s
7h ago254 ms80 tok/s
7h 15m ago233 ms88 tok/s
7h 30m ago234 ms84 tok/s
7h 45m ago246 ms85 tok/s
8h ago245 ms86 tok/s
8h 15m ago244 ms86 tok/s
8h 30m ago243 ms82 tok/s
8h 45m ago243 ms86 tok/s
9h ago232 ms86 tok/s
9h 15m ago239 ms86 tok/s
9h 30m ago249 ms82 tok/s
9h 45m ago312 ms75 tok/s
10h ago239 ms88 tok/s
10h 15m ago231 ms89 tok/s
10h 30m ago231 ms86 tok/s
10h 45m ago247 ms83 tok/s
11h ago246 ms86 tok/s
11h 15m ago226 ms87 tok/s
11h 30m ago235 ms87 tok/s
11h 45m ago240 ms84 tok/s
12h ago223 ms90 tok/s
12h 15m ago242 ms86 tok/s
12h 30m ago241 ms87 tok/s
12h 45m ago238 ms87 tok/s
13h ago222 ms91 tok/s
13h 15m ago243 ms84 tok/s
13h 30m ago223 ms88 tok/s
13h 45m ago239 ms83 tok/s
14h ago222 ms90 tok/s
14h 15m ago224 ms91 tok/s
14h 30m ago239 ms86 tok/s
14h 45m ago230 ms87 tok/s
15h ago234 ms86 tok/s
15h 15m ago229 ms89 tok/s
15h 30m ago233 ms87 tok/s
15h 45m ago229 ms90 tok/s
16h ago222 ms87 tok/s
16h 15m ago241 ms84 tok/s
16h 30m ago236 ms84 tok/s
16h 45m ago223 ms87 tok/s
17h ago237 ms86 tok/s
17h 15m ago234 ms87 tok/s
17h 30m ago230 ms87 tok/s
17h 45m ago238 ms84 tok/s
18h ago243 ms83 tok/s
18h 15m ago245 ms84 tok/s
18h 30m ago233 ms87 tok/s
18h 45m ago240 ms87 tok/s
19h ago230 ms87 tok/s
19h 15m ago240 ms86 tok/s
19h 30m ago231 ms89 tok/s
19h 45m ago225 ms87 tok/s
20h ago227 ms89 tok/s
20h 15m ago242 ms84 tok/s
20h 30m ago250 ms82 tok/s
20h 45m ago229 ms89 tok/s
21h ago240 ms86 tok/s
21h 15m ago235 ms84 tok/s
21h 30m ago238 ms85 tok/s
21h 45m ago239 ms88 tok/s
22h ago232 ms84 tok/s
22h 15m ago246 ms84 tok/s
22h 30m ago253 ms82 tok/s
22h 45m ago234 ms88 tok/s
23h ago246 ms86 tok/s
23h 15m ago249 ms83 tok/s
23h 30m ago247 ms83 tok/s
23h 45m ago235 ms87 tok/s
24h ago253 ms85 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": "qwen3-235b-a22b",
    "messages": [
      { "role": "user", "content": "What does a KV cache do?" }
    ]
  }'
POST /v1/chat/completionsapi.weir.example

Dedicated

Need reserved capacity?

Qwen3 235B A22B 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