Back to the models list Active on this gateway

Deepseek: deepseek-v4-flash-vision-exp

deepseek/deepseek-v4-flash-vision-exp Start building free

deepseek-v4-flash-vision-exp is an experimental multimodal model from DeepSeek, part of the broader DeepSeek-V4 family and tuned for very fast, cost-efficient inference. The "flash" naming generally indicates a latency-optimized, lighter-weight or more aggressively served variant, while "vision" signals full image understanding alongside text capabilities such as reasoning, coding, and content generation. It is intended for interactive applications that need quick responses on both language and vision tasks—like chatbots with image input, lightweight agents, and product integrations—rather than pushing absolute state-of-the-art accuracy at any cost. As an experimental tier, it is typically used to test new serving, efficiency, or architectural ideas before they are folded into more stable flagship releases.

Modalities
Context 1M tokens
Quota cost Standard (1×)
Added September 23, 2026

Pricing

Current price per 1,000,000 tokens. Every request is billed automatically from its exact usage.

Input $0,2587 / 1M tokens ≈ $0,00026 / 1K tokens
Output $0,7762 / 1M tokens ≈ $0,00078 / 1K tokens

Included in plans

This model is not included in any plan yet.

Statistics

Live performance and usage of this model, aggregated across all users.

Rolling 30 days Aggregated across all users · refreshed every few minutes
Requests 4 4 all time
Success rate 100% 0 failed requests
Median latency 2,76 s p95 2,93 s
Throughput 54,9 output tokens per second
Tokens processed 1,1K 269 avg. per request
Est. cost per request $0,00017 ≈ $0,1730 per 1K requests
Popularity #15 of 35 0,02% of all requests
Uptime (24h) — 0 requests

Usage over time

Token usage

Last 30 days
Input tokens
275 25,6%
Output tokens
800 74,4%
Avg. input per request
69
Avg. output per request
200
Largest request
684 tokens

Reliability

Last 30 days
Successful
4 100%
Failed
0 0%
  • 200 Success 4 100%

Response time

Successful requests · last 30 days
Fastest2,72 s
Average3,64 s
Median2,76 s
95th percentile2,93 s
99th percentile2,93 s
Slowest6,17 s

Last 24 hours

≥ 99% ≥ 90% < 90%
Latency (ms)
Requests per hour

Usage rhythm

Last 30 days · site timezone
  • Busiest hour: 12:00
  • Busiest weekday: Thursday
  • Busiest day: 19 November (4 requests)

All-time record

Total requests
4
Total tokens
1,1K
Active days
1
First request
19 November 2026
Last request
3 روز ago

API

Call this model from any OpenAI-compatible client using your API token.

curl
curl https://apiex.cloud/api/v1/chat/completions \
  -H "Authorization: Bearer $NEW_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "deepseek/deepseek-v4-flash-vision-exp",
  "messages": [
    {
      "role": "user",
      "content": "Explain quantum entanglement in one paragraph."
    }
  ],
  "temperature": 0.7
}'
Authentication

Every request must include an Authorization: Bearer <TOKEN> header. Generate a token from your account's API Keys page — its access to models is set automatically from your active plans and redeemed codes, and you can set its rate limit and IP restrictions.

Supported parameters
Parameter Type Description
modelstring, requiredThis model's ID, exactly as shown above.
messagesarray, requiredConversation history as role/content objects (system, user, assistant).
temperaturenumber, optionalSampling randomness, typically 0–2.
max_tokensinteger, optionalUpper limit on tokens generated in the reply.
top_pnumber, optionalNucleus-sampling threshold, as an alternative to temperature.
streamboolean, ignoredResponses are always returned in full so the exact cost can be billed; streaming is not available.

Replace YOUR_API_KEY with a token from your account, then see the full API reference for every parameter.

Frequently asked questions

What is deepseek-v4-flash-vision-exp?

deepseek-v4-flash-vision-exp is an experimental multimodal model from DeepSeek, part of the broader DeepSeek-V4 family and tuned for very fast, cost-efficient inference. The "flash" naming generally indicates a latency-optimized, lighter-weight or more aggressively served variant, while "vision" signals full image understanding alongside text capabilities such as reasoning, coding, and content generation. It is intended for interactive applications that need quick responses on both language and vision tasks—like chatbots with image input, lightweight agents, and product integrations—rather than pushing absolute state-of-the-art accuracy at any cost. As an experimental tier, it is typically used to test new serving, efficiency, or architectural ideas before they are folded into more stable flagship releases.

How much does deepseek-v4-flash-vision-exp cost?

Input tokens are priced at $0,2587 per 1,000,000 tokens and output tokens at $0,7762 per 1,000,000 tokens. Pricing is billed automatically per request; no separate plan is required.

What is the context length of deepseek-v4-flash-vision-exp?

deepseek-v4-flash-vision-exp supports a context window of up to 1,048,576 tokens.

Does deepseek-v4-flash-vision-exp support image input?

Yes, deepseek-v4-flash-vision-exp accepts image input in addition to text.

Which provider is deepseek-v4-flash-vision-exp from?

deepseek-v4-flash-vision-exp is developed by Deepseek and made available on this gateway through our routing infrastructure.

When was deepseek-v4-flash-vision-exp added to this gateway?

deepseek-v4-flash-vision-exp has been available on this gateway since September 23, 2026.

Is deepseek-v4-flash-vision-exp currently available?

Yes, deepseek-v4-flash-vision-exp is active and can be used right now through this gateway.

Explore more models

Keep exploring