All NewsEducationTV
Equities & FundsCrypto & Digital AssetsAI & TechnologyBusiness & CorporateUS Politics & PolicyGeopolitics & Global RiskMacro, Rates & FXCommodities & EnergyEuropean Politics & MarketsAsia-PacificReal Estate & Property
Story archiveAll categories
← All Stories

DeepSeek's V4-Flash AI model is cheapest to run among known models, research says

Created at 3 Aug · 5:44 AM1 source↑ Market-relevant
IN SHORT

A version of Chinese startup DeepSeek's V4-Flash AI model is significantly cheaper to run than competitors like Anthropic's Claude Fable 5 and OpenAI's GPT-5.6 Sol, according to research firm Artificial Analysis. The model achieved a score of 50 out of 100 on the firm's Intelligence Index.

✉Newsletter

PiQ Daily

Pick your topics. Get only what matters, on your cadence.

Key Numbers

$0.14DeepSeek V4-Flash cost per million input tokens
$0.28DeepSeek V4-Flash cost per million output tokens
3 centsV4-Flash average cost per test
86 centsKimi K3 average cost per test
$1.86GPT-5.6 Sol average cost per test
$3.15Claude Fable 5 average cost per test
50/100V4-Flash score on Intelligence Index
57Kimi K3 score on Intelligence Index

Who's Involved

DeepSeek
Chinese AI startup releasing V4-Flash model
Artificial Analysis
Research firm that benchmarked AI models
Anthropic
Company with Claude Fable 5 model
Moonshot AI
Chinese rival with Kimi K3 model
OpenAI
Company with GPT-5.6 Sol model
Google
Company with Gemini 3.6 Flash model
Meta
Company with Muse Spark 1.1 model
Z.AI
Chinese startup also known as Zhipu
Alibaba
Tech giant unveiling Qwen3.8-Max AI model

↳ Why This Matters

DeepSeek's focus on ultra-low-cost AI models could pressure larger tech companies to reduce their pricing, potentially accelerating AI adoption for businesses seeking to deploy the technology at scale. The research also highlights the increasing competitiveness of Chinese AI startups in the global market.

Key facts

  • DeepSeek's V4-Flash AI model is the cheapest to run among well-known global models, according to Artificial Analysis.
  • V4-Flash costs $0.14 per million input tokens and $0.28 per million output tokens.
  • Artificial Analysis estimated V4-Flash's average cost at 3 cents per test, compared to 86 cents for Moonshot AI's Kimi K3 and $1.86 for OpenAI's GPT-5.6 Sol.
  • The V4-Flash model scored 50 out of 100 on Artificial Analysis's Intelligence Index, matching Google's Gemini 3.6 Flash.
  • Alibaba has unveiled its Qwen3.8-Max AI model.

A version of Chinese startup DeepSeek's flagship AI model, V4-Flash, is significantly cheaper to run than its well-known global competitors, according to research firm Artificial Analysis. The model, officially released on Friday, charges $0.14 per million input tokens and $0.28 per million output tokens. Artificial Analysis estimated V4-Flash's average cost at 3 cents per test, a fraction of the cost for models from rivals such as Moonshot AI, OpenAI, and Anthropic.

This comparison accounts for the processing steps required to complete a task, offering a more realistic measure of value than headline pricing alone. DeepSeek's V4-Flash model achieved a score of 50 out of 100 on Artificial Analysis's Intelligence Index, which combines results from nine benchmarks. This score is equal to Google's Gemini 3.6 Flash and slightly behind Meta's Muse Spark 1.1 and Z.AI's GLM-5.2. However, Moonshot's Kimi K3 scored higher at 57, and Anthropic's Claude Opus 5 and OpenAI's GPT-5.6 scored even higher.

DeepSeek, known for offering ultra-low-cost AI alternatives, is also preparing a more powerful version, V4-Pro. The company was once a dominant force in Chinese AI development but now faces competition from numerous domestic startups and tech giants like ByteDance and Alibaba, all vying for global adoption by businesses seeking scalable and affordable AI solutions. Separately, Alibaba announced its largest and most capable AI model to date, Qwen3.8-Max.

Frequently asked questions

DeepSeek's V4-Flash is a new AI model released by the Chinese startup, designed to be ultra-low-cost to run.

According to Artificial Analysis, V4-Flash is significantly cheaper to run than models from Anthropic, OpenAI, and Moonshot AI, with an average cost of 3 cents per test.

The Intelligence Index is a benchmark developed by Artificial Analysis that combines results from nine tests covering coding, reasoning, and workplace assignments, scoring models out of 100.

The research included models such as Anthropic's Claude Fable 5 and Claude Opus 5, OpenAI's GPT-5.6 Sol, Moonshot AI's Kimi K3, Google's Gemini 3.6 Flash, Meta's Muse Spark 1.1, and Z.AI's GLM-5.2.

What Happens Next

01DeepSeek is preparing to release a more powerful version of its model, V4-Pro.

Get the newsletter.

Pick the topics you actually care about. We'll email when there's news worth your time, on the cadence you choose. Cancel any time from your account.

Cadence

How It Developed

DeepSeek officially released its V4-Flash AI model.
Research firm Artificial Analysis stated V4-Flash is the cheapest to run among well-known global AI models.
Artificial Analysis estimated V4-Flash's average cost at 3 cents per test, significantly lower than rivals.
DeepSeek's V4-Flash model scored 50 out of 100 on Artificial Analysis's Intelligence Index.
Alibaba unveiled its Qwen3.8-Max AI model.

Sources

T1
DeepSeek's new AI model is by far the cheapest of well-known models to run, research firm saysReuters

Related Stories

Alibaba's Qwen AI model performance questioned amid lack of benchmarks
3 Aug · 5:51 AM
US ban on Chinese AI models could cost businesses $12B annually: report
3 Aug · 3:10 AM
AI chip market volatility highlights opaque economy
2 Aug · 9:11 AM
China's Tech Giants Integrate AI Into Delivery Rider Helmets
3 Aug · 7:05 AM
Hugging Face CEO calls for mandatory AI hack disclosures
3 Aug · 7:00 AM