DeepSeek-V4-Flash
- In Input modalities: Text Out Output modalities: Text
- Reasoning· low, high, max
- Reasoning can be off
- Tool use
- Structured output
- Open weights
Prices
5 offerings · prices as listeddeepseek-v4-flash| Conditions/ 1M tokens | Input | Cache read | Output |
|---|---|---|---|
| Standard | ¥3.00 | ¥0.10 | ¥9.00 |
| Flex tier | ¥1.50 | ¥0.10 | ¥4.50 |
| Batch | ¥1.50 | ¥0.10 | ¥4.50 |
Prices from 火山方舟：模型价格 (opens in a new tab) (checked Nov 1, 2026)
deepseek-v4-flash| Conditions/ 1M tokens | Input | Cache read | Output |
|---|---|---|---|
| Standard no longer listed since Oct 11, 2026 | $0.138 | $0.028 | $0.275 |
Prices from Alibaba Cloud Model Studio pricing (opens in a new tab) (checked Oct 11, 2026) · deepseek-v4-flash model (opens in a new tab) (checked Oct 6, 2026)
deepseek-v4-flash| Conditions/ 1M tokens | Input | Cache read | Output |
|---|---|---|---|
| Standard no longer listed since Oct 11, 2026 | $0.20 | $0.040 | $0.40 |
Prices from Alibaba Cloud Model Studio pricing (opens in a new tab) (checked Oct 11, 2026) · deepseek-v4-flash model (opens in a new tab) (checked Oct 4, 2026)
deepseek-v4-flash| Conditions/ 1M tokens | Input | Output |
|---|---|---|
| Standard no longer listed since Oct 11, 2026 | ¥1.00 | ¥2.00 |
Prices from 阿里云百炼：模型价格 (opens in a new tab) (checked Oct 8, 2026)
Standard: the price a request with no conditions pays (the headline price elsewhere on the site).
Details
- Knowledge cutoff
- May 2025
- Licence
- MIT
- Family
- deepseek-flash
- Model ID
Not stated on DeepSeek’s pages: max input, max thinking budget, built-in tools.
Run locally
Weights: deepseek-ai/DeepSeek-V4-Flash- Parameters 290.9B
- Active 13B · 256 experts, 6 per token
- Native precision FP8
- Layers 43
- KV cache per token 86 KB
- Config positions 1024K (config.json max_position_embeddings)
- API context 977K (maker’s API)
- Needs
- ≈ 151.8 GB (weights 143.9 + KV cache 0.7 + overhead 7.2)
- Usable
- 21.6 GB (90% of its memory)
- Try
- IQ2_XXS runs (needs system memory), but below 4 bits quality drops noticeably
| Precision | Weights | Needs at 8K | On your hardware | Longest context |
|---|---|---|---|---|
| FP8 (native)maker’s file | 148.7 GB | ≈ 156.8 | ✕Doesn’t fit | — |
| Q6_K | 222.5 GB | ≈ 234.3 | ✕Doesn’t fit | — |
| Q5_K_M | 192.7 GB | ≈ 203.0 | ✕Doesn’t fit | — |
| Q5_0 | 186.3 GB | ≈ 196.3 | ✕Doesn’t fit | — |
| Q4_K_M | 164.6 GB | ≈ 173.5 | ✕Doesn’t fit | — |
| Q4_0 | 152.4 GB | ≈ 160.7 | ✕Doesn’t fit | — |
| IQ4_XS | 143.9 GB | ≈ 151.8 | ✕Doesn’t fit | — |
| Q3_K_Mquality drops | 132.4 GB | ≈ 139.7 | ✕Doesn’t fit | — |
| IQ3_XXSquality drops | 103.6 GB | ≈ 109.5 | ✕Doesn’t fit | — |
| Q2_Kquality drops | 101.6 GB | ≈ 107.4 | ✕Doesn’t fit | — |
| IQ2_XXSquality drops | 69.8 GB | ≈ 73.9 | ○Needs system memory 17–25 tok/s | — |
Click a row to choose a precision. “Maker’s file” sizes are the maker’s own files; others are parameters × bits per weight.
Compared with other hardware
Each at the highest precision it runs. Click one for details.
Smallest that runs it (Q4_K_M, 8K): no single listed GPU; unified memory Apple M5 Ultra (64-core GPU) · 256 GB.
Run commands
The maker publishes no files at this precision. Choose its native precision or one of its own builds, or quantise it yourself with llama.cpp.
Memory to fine-tune
Batch 1, 2,048-token sequences, gradient checkpointing; LoRA and QLoRA train rank-16 adapters on the attention projections.
Estimates; facts from the maker’s Hugging Face repositories, as of 2026-11-06. How it’s estimated
Cited scores
From third-party leaderboards, not measured by us- LMArena Text 1,436
#66 of 163 listed models · CI 1,432–1,440 · 51,944 votes
- EQ-Bench Creative Writing v3 1,559
#46 of 93 listed models
- EQ-Bench Longform Writing 66
#49 of 92 listed models
- LiveBench · Agentic Coding 37.6
#56 of 57 listed models
- LMArena (Arena), leaderboard-dataset on Hugging Face, licensed CC BY 4.0 · License · as of Oct 28, 2026
- EQ-Bench by Samuel J. Paech (eqbench.com; EQ-bench/EQ-bench-site on GitHub), MIT License. · License · as of Oct 28, 2026
- LiveBench (White et al.), release 2026-06-25, livebench.ai, licensed CC BY-SA 4.0. This board, adapted from it, is shared under CC BY-SA 4.0. · License · as of Jun 25, 2026
- changes we made
Included in coding plans
- 5× the usage of the Lite plan
Changes
All changes →- Removed from the directory
- Now priced on Alibaba Cloud Model Studio, China Input ¥1.00 · Output ¥2.00
- Now priced on Alibaba Cloud Model Studio, International Input $0.20 · Output $0.40 · Cache read $0.040
Sources
- deepseek-v4-flash model (opens in a new tab) alibabacloud.com · checked Oct 6, 2026
- DeepSeek (opens in a new tab) alibabacloud.com · checked Oct 6, 2026
- deepseek-v4-flash model (opens in a new tab) alibabacloud.com · checked Oct 9, 2026
- deepseek-v4-flash model (opens in a new tab) alibabacloud.com · checked Oct 4, 2026
- models.dev: deepseek-v4-flash (community data, MIT) (opens in a new tab) github.com · checked Sep 30, 2026
- Create a response (opens in a new tab) alibabacloud.com · checked Oct 9, 2026
- 火山方舟：模型价格in Chinese (opens in a new tab) docs.volcengine.com · checked Nov 1, 2026
- 硅基流动：模型价格in Chinese (opens in a new tab) siliconflow.cn · checked Oct 30, 2026
- 阿里云百炼：模型价格in Chinese (opens in a new tab) help.aliyun.com · checked Oct 8, 2026
- DeepSeek API change log (opens in a new tab) api-docs.deepseek.com · checked Nov 2, 2026
- 模型上下架与更新-大模型服务平台百炼-阿里云_大模型服务平台百炼(Model Studio)-大模型服务平台百炼(Model Studio)-阿里云帮助中心in Chinese (opens in a new tab) help.aliyun.com · checked Nov 4, 2026
- Explore Qwen3, Wan2.6 & Latest Models on Model Studio - Alibaba Cloud Model Studio - Alibaba Cloud_Alibaba Cloud Model Studio(Model Studio) - Alibaba Cloud Model Studio - Alibaba Cloud Documentation Center (opens in a new tab) alibabacloud.com · checked Nov 3, 2026
- Text generation models - Qwen, DeepSeek, GLM | Model Studio - Alibaba Cloud Model Studio - Alibaba Cloud Documentation Center (opens in a new tab) alibabacloud.com · checked Oct 9, 2026
- Alibaba Cloud Model Studio pricing (opens in a new tab) alibabacloud.com · checked Oct 11, 2026
- DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence (opens in a new tab) huggingface.co · checked Oct 31, 2026
- DeepSeek-V4-Flash-0731 (opens in a new tab) huggingface.co · checked Oct 30, 2026
- deepseek-ai/DeepSeek-V4-Flash model card (opens in a new tab) huggingface.co · checked Nov 6, 2026
- deepseek-ai/DeepSeek-V4-Flash config.json (opens in a new tab) huggingface.co · checked Oct 30, 2026
- deepseek-ai/DeepSeek-V4-Flash files (opens in a new tab) huggingface.co · checked Oct 31, 2026
- deepseek-ai/DeepSeek-V4-Flash README.md (opens in a new tab) huggingface.co · checked Nov 4, 2026
- DeepSeek V4 Preview Release | DeepSeek API Docs (opens in a new tab) api-docs.deepseek.com · checked Oct 31, 2026
- DeepSeek API models & pricing (opens in a new tab) api-docs.deepseek.com · checked Nov 2, 2026
DeepSeek’s official policies: data use, privacy, terms →
Some values on this page are still waiting to be checked against official pages: a filled dot is verified, a hollow one is not yet (with or without a source on file). How we verify data
DeepSeek · China · deepseek.com
Related models
Snapshots, the same family and other models from this provider