deepseek
DeepSeek V4 Flash Vision Exp
DeepSeek-V4-Flash-Vision-Exp is an experimental multimodal model that combines the agentic reasoning, coding, and world knowledge capabilities of DeepSeek-V4-Flash with advanced visual understanding. It delivers major gains on vision-dependent agent workflows while performing strongly across leading coding, repository, data science, automation, and multimodal benchmarks.
reasoningtool useimplicit cachingvisionstructured outputtextimage
Overview
- Type
- language
- Context
- 1M
- Max output
- 1M
- Input / output
- text, image / text
API
Model identifier
deepseek/deepseek-v4-flash-vision-expProviders 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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| deepinfra | Operational | 1.52s / 3.58s | 178.0 / 232.5 tok/s | 100.00% / 100.00% / 99.76% | $0.22/M | $0.65/M | $0.01/M / — | — | — | Yes | Yes |
| novita | Operational | — | — | — / — / — | $0.44/M | $1.32/M | $0.03/M / — | — | — | — | No |
Pricing
input
0.0000002156
output
0.0000006468
input cache read
0.0000000068
varies by provider
Yes
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",
"medium",
"high"
]
}
Metadata
type
language
context window
1M
max output
1M
released
Aug 21, 2026
temperature control
Yes
zero data retention
some
no training
some
api specifications
v2, v3, v4