DeepSeek: DeepSeek V3.1

deepseek/deepseek-chat-v3.1

DeepSeek-V3.1 is a large hybrid reasoning model (671B parameters, 37B active) that supports both thinking and non-thinking modes via prompt templates. It extends the DeepSeek-V3 base with a two-phase long-context...

Prompt $/1M

$0.250

Completion $/1M

$0.950

Context

163,840 tok

Best provider

53 tok/s · SambaNova

Specs

Model id: deepseek/deepseek-chat-v3.1
Display name: DeepSeek: DeepSeek V3.1
Modality: text->text
Tokenizer: DeepSeek
Instruct type: deepseek-v3.1
Context length: 163,840
Max completion: 32,768
Prompt $/1M: $0.250
Completion $/1M: $0.950
Free tier: no

Capabilities

Function callingno
JSON modeno
System promptyes
Visionno
Moderatedno

Benchmark scores

Each axis is one benchmark, normalized to the best model in the field.

No benchmark data for this model.

Speed by provider

Provider	tok/s	TTFT	Total latency
SambaNovafastest	53	1,056 ms	1,219 ms
WandB	52	357 ms	398 ms
Google	48	1,262 ms	1,420 ms
Together	35	716 ms	1,146 ms
AtlasCloud	29	1,171 ms	1,272 ms
Novita	20	1,607 ms	1,705 ms
Mara	20	6,309 ms	7,751 ms
SiliconFlow	18	1,274 ms	1,458 ms
DeepInfra	14	849 ms	985 ms
Chutes	11	1,782 ms	2,068 ms
Fireworks	6	1,278 ms	2,012 ms