As of: 2026-11-13 13:49 CET.
Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/
Canonical page: https://postcutoff.com/m/deepseek-v4-1-flash/

# DeepSeek-V4.1-Flash

DeepSeek, DeepSeek V4 family. Status: current. Type: reasoning model. Released: 2026-09-10. Training cutoff: not published.

## How to call it

| Provider | Model id | Endpoint | Docs |
|---|---|---|---|
| DeepSeek API | `deepseek-flash` | https://api.deepseek.com | https://api-docs.deepseek.com/quick_start/pricing |
| DeepSeek API (Anthropic format) | `deepseek-flash` | https://api.deepseek.com/anthropic | https://api-docs.deepseek.com/quick_start/pricing |
| Alibaba Cloud Model Studio | `deepseek-v4.1-flash` |  | https://www.alibabacloud.com/help/en/model-studio/models |
| OpenRouter | `deepseek/deepseek-v4.1-flash` |  | https://openrouter.ai/deepseek/deepseek-v4.1-flash |
| Hugging Face |  |  | https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash |
| Web app |  |  | https://chat.deepseek.com |

## Pricing

- Input: $0.30 per 1M tokens
- Output: $1.20 per 1M tokens
- Cache read: $0.006 per 1M tokens
- Unit as published: per 1M tokens (USD), peak-hour list price; off-peak is half (input 0.15, output 0.6, cache hit 0.003). Peak = 01:00-04:00 and 06:00-10:00 UTC Mon-Fri
- Source: https://api-docs.deepseek.com/quick_start/pricing
- Last checked: 2026-10-17

## Specs

- Input: text, image
- Output: text
- Context window: 1,000,000 tokens
- Max output: 384,000 tokens
- Open weights: yes
- Licence: mit

## What it has not seen

- No published training cutoff. Counted from its release (2026-09-10): at least 1,530 AI events since (329 major, 62 historic), as of 2026-11-13.
- Everything since August 2026: https://postcutoff.com/since/2026-08/

## What stands out

- Native vision in the Flash tier: First DeepSeek Flash model with native multimodal (image) understanding built in; replaced the separate V4-Flash-Vision-Exp. Source: https://api-docs.deepseek.com/updates
- Causal Encoder-Decoder (CED) architecture: 552B-backbone MoE that activates only ~8B params per token in prefill and ~16B in decode, aimed at input-heavy agentic workloads. Source: https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash
- Tiny KV cache (CSA2 + FP4 KV): Compressed Sparse Attention 2 and FP4 main KV cache cut the global KV cache to ~890 bytes/token, about 1/4 of V4-Flash. Source: https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash
- Hybrid thinking with effort levels: One model id serves thinking (default) and non-thinking modes; reasoning effort low/high/max. Source: https://api-docs.deepseek.com/updates
- Multiple API protocols: Same model served via OpenAI Chat Completions, OpenAI Responses (Codex-adapted) and Anthropic Messages formats. Source: https://api-docs.deepseek.com/quick_start/pricing

## Notes

Call as deepseek-flash (since Oct 17, 2026 it serves the DeepSeek-V4.1-Flash-1017 snapshot, same price; weights at huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017, OpenRouter id deepseek/deepseek-v4.1-flash-1017). Legacy ids deepseek-v4-flash and deepseek-v4-flash-vision-exp are routed here and billed at Flash price. Knowledge cutoff not published.

DeepSeek's cheapest current model: agentic coding, long-context (1M) work and image understanding at very low cost. Schedule batch jobs off-peak for 50% off.

```bash
curl https://api.deepseek.com/chat/completions \
 -H "Authorization: Bearer $DEEPSEEK_API_KEY" -H "Content-Type: application/json" \
 -d '{"model":"deepseek-flash","messages":[{"role":"user","content":"Hello"}]}'
```

Sources: https://api-docs.deepseek.com/quick_start/pricing · https://api-docs.deepseek.com/updates · https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash

Model page: https://postcutoff.com/m/deepseek-v4-1-flash/