DeepSeek-V4.1-Flash
deepseek-v4-1-flashLong-context reasoning, multimodal understanding and tool calling.
- Modality
- Text
- List price
- Input $0.148 · Output $0.594 / million tokensCache read $0.003 / million tokens
- Context window
- 1,024K
Model specifications and parameters
Fields are shown for this model's connected endpoint, not one generic form for every model.
- Context window
- 1,024K
- Maximum input
- 1,024K
- Provider maximum output
- 384K
- Maximum reasoning content
- 384K
| Parameter / field | Current range | Description |
|---|---|---|
| messages | Required | Submit ordered conversation messages with their roles. |
| max_completion_tokens | 1–8,192 (web test) | The web test uses a lower cost-estimate guardrail; the provider limit appears above. |
| temperature | 0–2 | Sampling temperature: higher is more varied, lower is more focused. |
| top_p | 0.95–1 (thinking) / 1 (off) | Nucleus sampling; V4.1 Flash's effective range changes with thinking mode. |
| stream | true / false | Return chunks as generation progresses. |
| thinking.type | enabled / disabled | Support and value mapping vary by model version; the web experience shows only verified controls. |
| reasoning_effort | minimal / low / medium / high / xhigh / max | Reasoning levels available for this version; adjacent levels may map to the same effective setting. |
Provider limits and LINK42 web-test limits may differ. LINK42's current quote and actual settlement govern price; external docs are provided to verify model capabilities.
Test a result before integrating
Adjust parameters, review the estimate and validate your idea with an actual output.
Test parameters
deepseek-v4-1-flashChecking your session…
Your first result starts here
Enter a prompt and start a test. Generated text, images or video will appear here.
Test history
Private to your account, including recent results and actual charges.
Explore the model and adjust parameters first. Sign in to restore your draft and run a paid test.
Overview
DeepSeek-V4.1-Flash
Capabilities depend on the selected model and its current configuration. Review pricing, then validate a small sample with your own task.
Frequently asked questions
Are playground tests billed?
Yes. Tests use the signed-in account's balance. The estimated hold is shown before submission, and the final amount is calculated from actual usage and effective pricing.
What happens if a test fails or stops?
Requests rejected before generation have no model usage. Failures, cancellations and interrupted streams are settled or released by the shared ledger. Check the run record for any billable usage already produced.
Where can I see results and charges?
Recent results and actual charges are available in the test history. Open the usage record using its request ID. Download large results promptly.
Pricing
These are standard model prices. Signed-in estimates account for your billing identity and parameters; final charges use actual usage.
- Input price
- $0.148476/ million tokens
- Output price
- $0.593904/ million tokens
- Cache reads
- $0.00297/ million tokens
API
After validating the result, use the same model identifier in your application. Keep API keys on your server.
curl https://www.link42.ai/v1/chat/completions -H "Authorization: Bearer $LINK42_API_KEY" -H "Content-Type: application/json" -d '{"model": "deepseek-v4-1-flash", "messages": [{"role": "user", "content": "…"}]}'