DeepSeek: DeepSeek V4 Flash 0423

deepseek/deepseek-v4-flash

DeepSeek V4 Flash 0423 (DeepSeek) is a low-cost open-weight model priced near the bottom of the board, at $0.091/$0.182 per million input/output tokens, served by 13 independent hosts. For the money it delivers #62 of 260 on the TryAii Score — among the best score-per-dollar in its class. Its best result is knowledge (#32 on GPQA). It's reasonably quick at roughly 33-69 tokens/sec.

Prompt $/1M

$0.091

Completion $/1M

$0.182

Context

1,048,576 tok

Best provider

85 tok/s · Baidu

Specs

Model id: deepseek/deepseek-v4-flash
Display name: DeepSeek: DeepSeek V4 Flash 0423
Modality: text->text
Tokenizer: DeepSeek
Instruct type: —
Context length: 1,048,576
Max completion: 943,717
Prompt $/1M: $0.091
Completion $/1M: $0.182
Free tier: no

Capabilities

Function callingno
JSON modeno
System promptno
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
Baidufastest	85	787 ms	1,052 ms
Azure	83	1,182 ms	1,395 ms
Relace	78	486 ms	650 ms
Alibaba	74	1,141 ms	1,245 ms
Cloudflare	67	1,705 ms	2,321 ms
Wafer	65	999 ms	1,343 ms
Parasail	65	634 ms	769 ms
GMICloud	50	1,793 ms	1,908 ms
AtlasCloud	47	2,162 ms	3,051 ms
StreamLake	42	1,251 ms	1,640 ms
SiliconFlow	38	908 ms	1,036 ms
Mancer 2	33	1,012 ms	1,325 ms
DeepInfra	32	999 ms	1,259 ms
DigitalOcean	28	632 ms	766 ms
Venice	24	1,591 ms	1,884 ms
OpenInference	12	1,868 ms	2,532 ms