DeepSeek-V4-Flash-0731
DeepSeek-V4-Flash-0731 is the official release of DeepSeek's V4-Flash (MIT licence, July 2026), replacing the April preview: a mixture-of-experts model with 284B parameters and 13B active per token, built for million-token context. DeepSeek reports substantially stronger agentic behaviour — code agents and tool use — and publishes Terminal Bench 2.1 at 82.7 on its card. Reasoning effort is selectable: low, high or max. It is fulfilled through global providers under our European contract, and requests go only to backends that do not retain prompts or completions.
model id: deepseek-v4-flash-0731
from openai import OpenAI client = OpenAI( api_key="hb-...", base_url="https://api.heabsy.com/v1",) r = client.chat.completions.create( model="deepseek-v4-flash-0731", messages=[{"role": "user", "content": "Hello"}],)print(r.choices[0].message.content)
curl https://api.heabsy.com/v1/chat/completions \ -H "Authorization: Bearer $HEABSY_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "deepseek-v4-flash-0731", "messages": [{"role": "user", "content": "Hello"}] }'
import OpenAI from "openai"; const client = new OpenAI({ apiKey: process.env.HEABSY_API_KEY, baseURL: "https://api.heabsy.com/v1",}); const r = await client.chat.completions.create({ model: "deepseek-v4-flash-0731", messages: [{ role: "user", content: "Hello" }],});console.log(r.choices[0].message.content);
Price list, updated 12 September 2026. No minimum spend, no subscription.
About the model
- Developer
- DeepSeek
- Released
- July 2026
- Licence
- MIT
- Architecture
- mixture of experts · 284B parameters, 13B active per token
- Input
- text
- Reasoning
- selectable effort
- Languages named by the developer
- none named on the model card
Source: the developer's model card, as of 17 September 2026.
Which DeepSeek model to choose
DeepSeek V4 Flash is the April 2026 preview. DeepSeek-V4-Flash-0731 is its official release (284B parameters, 13B active, text only) and the sensible default for code and tool agents at moderate cost.
DeepSeek-V4-Pro-0813 is the flagship: 1.6T parameters with 49B active. DeepSeek recommends allowing up to 384K output tokens at high and max effort — our maximum output for this model is listed under technical details.
DeepSeek-V4.1-Flash (September 2026) is a different network, not a point release: 552B parameters, 8B active while reading and 16B while writing, and it accepts images. Take it for input-heavy agents; keep the preview only if you need to reproduce earlier results.
All four are open weights under the MIT licence.
Licence: what it means for your company
MIT. A permissive open-source licence: you may use, modify and sell products built on the model, including commercially; when you redistribute the weights, the licence and copyright notices go with them.
A summary of the licence, not legal advice.
Benchmarks reported by the developer
- Terminal Bench 2.1
- 82.7
- DeepSWE
- 54.4
Figures from the developer's model card, not our measurement. Source: huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731
Technical details
- Context window
- 1.3M tokens
- Max output
- 65,536 tokens
- Throughput limit
- 200K tok/min
- Tokenizer
- DeepSeek
- Streaming
- yes
- Open weights
- deepseek-ai/DeepSeek-V4-Flash-0731
- max_tokens
- 1 … 65536
- temperature
- 0 … 2
- top_p
- 0 … 1
- frequency_penalty
- -2 … 2
- presence_penalty
- -2 … 2
- stop
- array
- seed
- integer
This model is fulfilled through global providers under our European contract: one DPA with an EU company, one invoice for every model in the catalog, no US counterparty on your paperwork. Compute may run outside the EEA — if you need strict EEA-only processing, use our own-hardware model.
Questions about this model
›Where is DeepSeek-V4-Flash-0731 computed?
Outside the EEA, through a global provider under our European contract. You get one DPA with an EU company and one invoice, but the compute itself does not run in the EEA — we state that instead of hiding it.
›Does DeepSeek-V4-Flash-0731 store prompts and completions?
No. This model runs with zero data retention: request content is not written to durable storage and nothing is used for training.
›How large is the DeepSeek-V4-Flash-0731 context window?
1.3M tokens, with up to 65,536 tokens of output.
›What does DeepSeek-V4-Flash-0731 cost?
$0.06 per million input tokens and $0.12 per million output tokens. No minimum spend and no subscription; the same price list covers the API and the invoice.
›Is there a throughput limit on DeepSeek-V4-Flash-0731?
Yes, 200K tokens per minute. Higher limits are a question of volume, not of principle.
›Which languages does DeepSeek-V4-Flash-0731 support?
The developer's model card names no languages. Test the model on your own material before relying on it in a particular language.
›Under which licence is DeepSeek-V4-Flash-0731 released?
MIT. A permissive open-source licence: you may use, modify and sell products built on the model, including commercially; when you redistribute the weights, the licence and copyright notices go with them.
›Can reasoning be switched off in DeepSeek-V4-Flash-0731?
The developer lets you choose how much the model reasons; the card does not document turning reasoning off completely.
Compare with similar models
| Model | Developer | Parameters | Context | Licence | Input / 1M |
|---|---|---|---|---|---|
| DeepSeek-V4-Flash-0731 | DeepSeek | 284B / 13B | 1.3M | MIT | $0.06 |
| DeepSeek V4 Flash | DeepSeek | 284B / 13B | 1M | MIT | $0.25 |
| DeepSeek-V4.1-Flash | DeepSeek | 552B / 8–16 | 1M | MIT | $0.15 |
| DeepSeek-V4-Pro-0813 | DeepSeek | 1.6T / 49B | 1M | MIT | $0.80 |
30 more models behind the same key and the same invoice.
Browse the catalog