deepseek
DeepSeek V4.1 Flash
DeepSeek V4.1 Flash is an AI model from DeepSeek designed for fast, efficient multimodal interactions. It combines native multimodal capabilities with a new architecture focused on stronger performance, faster responses, and lower computational cost.
reasoningtool useimplicit cachingvisionstructured outputtextimage
Overview
- Type
- language
- Context
- 1M
- Max output
- 33K
- Input / output
- text, image / text
API
Model identifier
deepseek/deepseek-v4.1-flashProviders and live performance
Rolling Gateway metrics. Latency is time to first token; throughput is output tokens per second.
| Provider | Status | Latency | Throughput | Uptime 15m / 1h / 1d | Input | Output | Cache read / write | Discount | Regions | ZDR | No training |
|---|---|---|---|---|---|---|---|---|---|---|---|
| alibaba | Operational | 3.24s / 14.87s | 234.5 / 237.0 tok/s | 100.00% / 100.00% / 99.90% | $0.15/M | $0.60/M | $0.01/M / — | — | — | Yes | Yes |
| baseten | Operational | 0.54s / 1.81s | 256.0 / 276.7 tok/s | 100.00% / 99.94% / 99.90% | $0.30/M | $1.20/M | $0.01/M / — | — | us | Yes | Yes |
| boundless | Operational | 0.94s / 3.80s | 611.0 / 632.6 tok/s | 100.00% / 100.00% / 100.00% | $0.20/M | $1.00/M | $0.01/M / — | — | — | No | No |
| deepinfra | Operational | 1.07s / 3.11s | 65.5 / 78.8 tok/s | 100.00% / 99.99% / 99.98% | $0.14/M | $0.42/M | $0.00/M / — | — | — | Yes | Yes |
| deepseek | Operational | 1.11s / 2.11s | 186.0 / 223.9 tok/s | 99.98% / 99.98% / 99.95% | $0.15/M | $0.60/M | $0.00/M / — | — | — | — | — |
| fireworks | Operational | 1.90s / 7.48s | 129.0 / 163.5 tok/s | 100.00% / 99.94% / 99.99% | $0.30/M | $1.20/M | $0.01/M / — | — | us | Yes | Yes |
| gmicloud | Operational | 0.88s / 1.35s | 209.5 / 268.6 tok/s | 100.00% / 100.00% / 98.56% | $0.18/M | $0.72/M | $0.00/M / — | — | — | No | No |
| inference-net | Operational | 0.34s / 1.05s | 125.0 / 147.1 tok/s | 100.00% / 100.00% / 100.00% | $0.30/M | $1.00/M | $0.01/M / — | — | — | No | No |
| modal | Operational | 1.03s / 8.43s | 263.0 / 378.0 tok/s | 100.00% / 100.00% / 99.99% | $0.30/M | $1.20/M | $0.03/M / — | — | — | Yes | Yes |
| morph | Operational | 0.97s / 5.47s | 44.0 / 81.8 tok/s | 100.00% / 100.00% / 100.00% | $0.04/M | $0.70/M | $0.01/M / — | — | — | Yes | Yes |
| novita | Operational | 1.38s / 14.65s | 212.0 / 235.8 tok/s | 100.00% / 100.00% / 100.00% | $0.20/M | $0.78/M | $0.00/M / — | — | — | — | No |
| parasail | Operational | 1.16s / 2.60s | 156.5 / 344.1 tok/s | 100.00% / 100.00% / 99.99% | $0.30/M | $1.20/M | $0.01/M / — | — | — | Yes | Yes |
| particle | Operational | 1.31s / 2.85s | 144.0 / 144.0 tok/s | 100.00% / 100.00% / 99.99% | $0.07/M | $1.00/M | $0.03/M / — | — | — | Yes | Yes |
| relace | Operational | 0.27s / 0.68s | 185.0 / 185.0 tok/s | 100.00% / 100.00% / 99.99% | $0.20/M | $0.60/M | $0.01/M / — | — | — | Yes | Yes |
| runinfra | Operational | 0.41s / 1.64s | 528.0 / 531.6 tok/s | 100.00% / 100.00% / 100.00% | $0.14/M | $0.58/M | $0.03/M / — | — | — | No | No |
| runware | Operational | 1.26s / 2.80s | 159.5 / 243.7 tok/s | 100.00% / 100.00% / 99.90% | $0.15/M | $0.60/M | $0.01/M / — | — | — | Yes | Yes |
| togetherai | Operational | 0.36s / 2.35s | 256.0 / 434.0 tok/s | 100.00% / 100.00% / 99.99% | $0.30/M | $1.20/M | $0.01/M / — | — | — | Yes | Yes |
| wafer | Operational | 1.45s / 11.55s | 100.0 / 106.3 tok/s | 100.00% / 100.00% / 100.00% | $0.10/M | $0.65/M | $0.06/M / — | — | — | Yes | Yes |
Pricing
input
0.0000003
output
0.0000012
input cache read
0.000000007
varies by provider
Yes
fast
{
"input": "0.0000006",
"output": "0.0000024",
"input_cache_read": "0.000000014"
}
regional
{
"us": {
"input": "0.0000003",
"output": "0.0000012",
"input_cache_read": "0.000000007"
}
}
Capabilities
reasoningtool useimplicit cachingvisionstructured outputtextimage
Supported parameters
max tokens
Yes
temperature
Yes
stop
Yes
tools
Yes
tool choice
Yes
reasoning
Yes
include reasoning
Yes
response format
Yes
structured outputs
Yes
Reasoning options
options
{
"type": "toggle"
}, {
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
Metadata
type
language
context window
1M
max output
33K
released
Sep 8, 2026
temperature control
Yes
zero data retention
some
no training
some
regions
us
api specifications
v2, v3, v4