Skip to content
Get an API key

Live · serverless · 2 regions

DeepSeek V3.1

by DeepSeek

A 671B mixture-of-experts model for general chat and code, with a hybrid mode that can reason before it answers.

  • chat
  • code
  • FP8
Serverless priceUSD
Input
$0.55per 1M tokens
Output
$1.65per 1M tokens

Billed per token, no minimums.

Reference

Specifications

Model ID
deepseek-v3-1
Endpoint
POST /v1/chat/completions
Publisher
DeepSeek
Tasks
chat · code
Served context
128K tokens
Quantization
FP8
Decode speed
62 tok/smedian, one request
p50 first token
280 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

280 msmedian

p95 367 ms

Decode speed · tok/s

62 tok/smedian

p5 53 tok/s

Show the data
Median of each 15-minute step over the last 24 hours, newest first
StepTime to first tokenDecode speed
15m ago288 ms61 tok/s
30m ago293 ms63 tok/s
45m ago298 ms59 tok/s
1h ago278 ms64 tok/s
1h 15m ago282 ms64 tok/s
1h 30m ago279 ms61 tok/s
1h 45m ago285 ms61 tok/s
2h ago302 ms60 tok/s
2h 15m ago295 ms61 tok/s
2h 30m ago296 ms60 tok/s
2h 45m ago303 ms60 tok/s
3h ago396 ms48 tok/s
3h 15m ago282 ms63 tok/s
3h 30m ago291 ms60 tok/s
3h 45m ago295 ms61 tok/s
4h ago298 ms59 tok/s
4h 15m ago279 ms62 tok/s
4h 30m ago291 ms60 tok/s
4h 45m ago279 ms62 tok/s
5h ago293 ms61 tok/s
5h 15m ago293 ms62 tok/s
5h 30m ago396 ms49 tok/s
5h 45m ago296 ms59 tok/s
6h ago282 ms64 tok/s
6h 15m ago293 ms59 tok/s
6h 30m ago279 ms63 tok/s
6h 45m ago277 ms61 tok/s
7h ago295 ms62 tok/s
7h 15m ago278 ms64 tok/s
7h 30m ago296 ms60 tok/s
7h 45m ago374 ms51 tok/s
8h ago295 ms60 tok/s
8h 15m ago284 ms63 tok/s
8h 30m ago282 ms62 tok/s
8h 45m ago268 ms64 tok/s
9h ago275 ms62 tok/s
9h 15m ago275 ms62 tok/s
9h 30m ago267 ms62 tok/s
9h 45m ago269 ms64 tok/s
10h ago277 ms62 tok/s
10h 15m ago263 ms64 tok/s
10h 30m ago289 ms62 tok/s
10h 45m ago269 ms65 tok/s
11h ago262 ms65 tok/s
11h 15m ago286 ms63 tok/s
11h 30m ago269 ms64 tok/s
11h 45m ago278 ms61 tok/s
12h ago282 ms61 tok/s
12h 15m ago279 ms63 tok/s
12h 30m ago272 ms61 tok/s
12h 45m ago261 ms65 tok/s
13h ago275 ms63 tok/s
13h 15m ago275 ms65 tok/s
13h 30m ago265 ms64 tok/s
13h 45m ago273 ms65 tok/s
14h ago277 ms62 tok/s
14h 15m ago280 ms61 tok/s
14h 30m ago281 ms62 tok/s
14h 45m ago275 ms61 tok/s
15h ago277 ms62 tok/s
15h 15m ago267 ms64 tok/s
15h 30m ago256 ms66 tok/s
15h 45m ago280 ms63 tok/s
16h ago270 ms64 tok/s
16h 15m ago266 ms65 tok/s
16h 30m ago257 ms64 tok/s
16h 45m ago277 ms63 tok/s
17h ago269 ms64 tok/s
17h 15m ago269 ms65 tok/s
17h 30m ago367 ms54 tok/s
17h 45m ago272 ms62 tok/s
18h ago275 ms64 tok/s
18h 15m ago279 ms61 tok/s
18h 30m ago279 ms63 tok/s
18h 45m ago269 ms62 tok/s
19h ago265 ms63 tok/s
19h 15m ago280 ms62 tok/s
19h 30m ago262 ms65 tok/s
19h 45m ago380 ms49 tok/s
20h ago276 ms64 tok/s
20h 15m ago284 ms63 tok/s
20h 30m ago282 ms62 tok/s
20h 45m ago278 ms62 tok/s
21h ago271 ms63 tok/s
21h 15m ago283 ms62 tok/s
21h 30m ago282 ms61 tok/s
21h 45m ago272 ms62 tok/s
22h ago286 ms63 tok/s
22h 15m ago366 ms53 tok/s
22h 30m ago346 ms55 tok/s
22h 45m ago287 ms62 tok/s
23h ago299 ms61 tok/s
23h 15m ago296 ms60 tok/s
23h 30m ago291 ms61 tok/s
23h 45m ago292 ms59 tok/s
24h ago280 ms64 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": "deepseek-v3-1",
    "messages": [
      { "role": "user", "content": "What does a KV cache do?" }
    ]
  }'
POST /v1/chat/completionsapi.weir.example

Dedicated

Need reserved capacity?

DeepSeek V3.1 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