DeepSeek V4 Flash

Retired Sep 2026

DeepSeek's lightweight 284B MoE (13B active) under the MIT license, previewed April 2026 and released as DeepSeek-V4-Flash-0731 in July 2026 with gains in agentic and coding work.

Params: 284B, 13B active
Context: 1M tokens
Max output: 384K
Released: Apr 2026
License: MIT

Visit DeepSeek Compare vs other LLMs →

Hosted API price

Cheapest via DigitalOcean · per 1M tokens

USD

Input: $0.08 /1M
Output: $0.25 /1M
Hosted by: 10 providers

Hosted API pricing

How we price

Order

Providers are listed cheapest input rate first, then by output rate, a tie going to the creator's own listing. A provider that lists the model without a published rate sits at the end.

Prices

Base-tier, on-demand rates in USD per 1M tokens, converted at ECB reference rates where a provider publishes in another currency. Cached-input and batch rates show only where a provider publishes them. "Cheapest" marks the one provider with the lowest input and output pair; a tie wears no badge.

Cost

Input rate times the input tokens plus output rate times the output tokens, at the monthly volume set above the table. Cached-input, batch, long-context and reasoning-token billing are not modelled.

Transparency and funding

Affiliates: Affiliate links are marked. We may earn a commission if you click them, but commissions never affect the order.

Every provider serving DeepSeek V4 Flash, cheapest input first. Set your monthly volume to see what each would bill.

Input M Output M

Every provider serving DeepSeek V4 Flash, with its price per 1M input and output tokens and what the volume above would cost
Provider	Input /1M	Output /1M	Cached input /1M	Cost at 10M in + 2M out	Link
DigitalOcean Cheapest	$0.08	$0.25	$0.03	$1.30	Visit website
EmpirioLabs AI	$0.14	$0.28		$1.96	Visit website
Novita	$0.14	$0.28	$0.03	$1.96	Visit website
Together AI	$0.14	$0.28	$0.03	$1.96	Visit website
Geodd	$0.14	$0.30		$2.00	Visit website
GPUhub	$0.15	$0.30		$2.10	Visit website
Lyceum	$0.15	$0.30	$0.04	$2.10	Visit website
Azure	$0.19	$0.51	$0.03	$2.92	Visit website
Cloudflare	$0.44	$1.32	$0.01	$7.04	Visit website
EcoHash	$0.49	$1.42	$0.05	$7.74	Visit website

Heads up: Base-tier, on-demand rates per 1M tokens; cached, batch and long-context tiers excluded. A provider may serve a shorter context or a quantized build than the creator's release. Verify before provisioning. More on how we price.

Estimated cost to self-host

DeepSeek V4 Flash needs about 158 GB of GPU memory at 4-bit with 32K context. The cheapest rental that fits is 8× RTX 3090 at about $1,267 a month.

That costs the same as roughly 12B tokens a month on DigitalOcean's API. Self-hosting is more expensive below that volume.

Single node ≤ 8

What it costs to self-host DeepSeek V4 Flash at each precision: the memory it needs, the cheapest rental that holds it, that rental's month and the API volume it pays off at
Precision	Memory	Cheapest fit (32K context)(Full 1M context)	Cost /mo	Break-even vs API
4-bit INT4 / FP4	158 GB164 GB	8× RTX 3090	$1,267$1,267	12B tokens /mo12B tokens /mo
8-bit FP8 / INT8	286 GB292 GB	8× A40	$2,822$2,822	26B tokens /mo26B tokens /mo
16-bit FP16 / BF16	570 GB576 GB	8× A100	$10,138$12,730	94B tokens /mo118B tokens /mo

Estimates based on median on-demand rates for Nvidia GPUs. Memory is weights plus KV cache for one request, using FP8 KV cache where supported and FP16/BF16 otherwise. Break-even assumes a 5:1 input-to-output ratio. No guarantee of runtime support, usable performance, or that a matching quantized build exists. How we estimate costs.

Similarly priced models

The models nearest DeepSeek V4 Flash by blended rate, each at its own cheapest provider.

Models priced nearest DeepSeek V4 Flash per 1M tokens, each at its cheapest provider
Model	Blended /1M	Input /1M	Output /1M	Context	Cutoff	vs DeepSeek V4 Flash
Llama 3 8B Meta	$0.0833	$0.05	$0.25	8K	Mar 2023	−23%
Qwen3-Coder-30B-A3B Alibaba Cloud	$0.0917	$0.06	$0.25	262K		−15%
Llama 3.2 3B Meta	$0.0983	$0.05	$0.34	128K	Dec 2023	−9%
Qwen3-30B-A3B FP8 Alibaba Cloud	$0.0983	$0.05	$0.34	33K		−9%
Ministral 3 3B Mistral	$0.10	$0.10	$0.10	256K		−8%
DeepSeek V4 Flash This model DeepSeek	$0.1083	$0.08	$0.25	1M
GLM-5.3-Flash Z.AI	$0.1083	$0.08	$0.25	1M		0%
GPT-5 Nano OpenAI	$0.1083	$0.05	$0.40	400K	May 2024	0%
Qwen3.5-35B-A3B Alibaba Cloud	$0.1267	$0.06	$0.46	262K		+17%
Qwen3.6-35B-A3B Alibaba Cloud	$0.1283	$0.07	$0.42	262K		+18%
Gemma 3 27B Instruct Google Cloud	$0.1333	$0.10	$0.30	128K	Aug 2024	+23%

USD per 1M tokens at each model's cheapest listed provider. Blended is the cost of 10M input plus 2M output tokens, spread over the 12M.

Runpod

Spin Up a GPU in Seconds - No Setup Required

Deploy AI Models Globally in 30+ Regions

Scale from 0 to 1000 GPUs on Demand

PyTorch, Tensorflow & Docker Ready

Learn more

Capabilities

What DeepSeek V4 Flash accepts and can do, as published by DeepSeek.

Text Accepts and generates natural-language text.
Reasoning Thinks before it answers, always on or as a switchable mode.
Function calling Connect to external tools, APIs, and systems.
Structured output Return responses in structured formats like JSON.
Web search Search the web for up-to-date information.

Common questions

What is DeepSeek V4 Flash good for?

Long-context agent and coding work. MIT-licensed weights, far lighter to self-host than V4 Pro, though still a multi-GPU node.

When is DeepSeek V4 Flash not a good fit?

New work: DeepSeek retired it and now routes the model name to V4.1 Flash. It never took image input, and self-hosting the weights takes a multi-GPU node.

What is the cheapest way to run DeepSeek V4 Flash?

Hosted, unless you push serious volume. DigitalOcean charges $0.08 in / $0.25 out per 1M tokens. The cheapest rental that fits is 8x RTX 3090 at $1,267 a month, which costs the same as about 12B tokens a month on that API.

How does DeepSeek V4 Flash compare with DeepSeek V4.1 Flash?

DeepSeek says V4.1 Flash moves the Flash line onto a Causal Encoder-Decoder architecture that activates fewer parameters on input than on output, adds native image understanding, and compresses the KV cache, which it says cuts the cache-hit charges that dominate agent costs. Same 1M-token context. DeepSeek V4 Flash costs 73% less per 1M input tokens and 79% less per 1M output tokens than DeepSeek V4.1 Flash, at each model's cheapest listed provider. It came out 5 months earlier, has fewer parameters (284B with 13B active against 552B with 16B active) and takes no image input, which DeepSeek V4.1 Flash does.

Can I self-host DeepSeek V4 Flash?

Yes. The weights are MIT licensed. At 4-bit it needs about 158 GB of GPU memory, which starts at roughly $1,267 a month on the cheapest rental that fits.