DeepSeek vs ChatGPT vs Claude: The $0 vs $20/Month vs API Decision
DeepSeek's web chat costs $0 with no message cap. ChatGPT Plus and Claude Pro are $20 a month each, and that $20 buys features, not better everyday text answers: voice and images on ChatGPT, Claude Code on Claude. For text-only personal use, start with DeepSeek. If you are building on the API, compare token rates instead: DeepSeek V4.1 Flash is $0.30/$1.20 per 1M tokens at peak, against $2/$10 on Claude Sonnet 5.
By Aniruddh PanvalkarPricing verified Nov 5, 2026
At a glance
| Plan | ChatGPT PlusOpenAI · Subscription | Claude ProAnthropic · Subscription |
|---|---|---|
| Monthly price | $20/month (USD) | $20/month (USD) |
| Included usage | Not published | At least 5x Free usage per five-hour session plus weekly limits; includes Claude Code, Opus, Projects, Research, Chrome and Microsoft 365 integrations. |
| Overage | Not published | Not published |
| Models available | GPT-4o | Not published |
| AUD price incl. GST | $31.36/monthestimated (ECB rate 2026-11-04, +10% GST) | $31.36/monthestimated (ECB rate 2026-11-04, +10% GST) |
| Verified | ||
| Source | openai.com | anthropic.com |
Quick verdict
Text only: DeepSeek, $0. Voice and images: ChatGPT Plus, $20. Daily coding: Claude Pro, $20 with Claude Code. Building on the API: test DeepSeek V4.1 Flash first.
Summary
DeepSeek's price is $0, and its costs are elsewhere: 'Server Busy' throttling at China daytime peak hours, and no voice mode, image generation or coding agent. Paying $20 makes sense only if you will use one of those missing features. On the API the gap is bigger than $20. A month of 10M input and 2M output tokens costs $5.40 on DeepSeek V4.1 Flash at peak rates, $40 on Claude Sonnet 5 and $110 on GPT-5.5.
Quick Decision
Text only: DeepSeek, $0. Voice and images: ChatGPT Plus, $20. Daily coding: Claude Pro, $20 with Claude Code. Building on the API: test DeepSeek V4.1 Flash first.
Cheapest option: DeepSeek V4 Flash
DeepSeek V4.1 Flash is $0.30 per 1M input tokens and $1.20 per 1M output at peak, the cheapest API on this page. It replaced V4 Flash; the API name is now deepseek-flash. V4 Pro at $1.32/$3.96 is the step up if Flash misses your quality bar. Test both on your own prompts before you move production traffic.
Pricing Comparison
ChatGPT Plus
OpenAI
$20/month
People who use voice mode, image generation, or GPTs and connectors, none of which DeepSeek's free chat matches.
Claude Pro
Anthropic
$20/month
Developers who code daily and want Claude Code included, plus long-document work on Claude Sonnet 5 and Opus 5.
Pricing and plans verified
Chat plans and API prices are separate bills
DeepSeek chat, ChatGPT Plus and Claude Pro are all consumer chat products. None of them includes API access. API use is billed per token on its own account, even if you already pay $20 for the app. So this page answers two questions. For your own use, compare $0 against $20. For a product, ignore the plan prices and compare token rates.
What $0 gets you, and what $20 adds
DeepSeek (free): no message cap, web search, file uploads and saved chat history, with no card needed. The limit is fair-use throttling, with 'Server Busy' warnings at China daytime peak hours. ChatGPT Plus ($20 a month): higher GPT-5.5 limits, voice mode, image generation, the largest GPTs and connector ecosystem, and the most mature mobile app. ChatGPT Go at $8 is a cheaper paid step. Claude Pro ($20 a month): Claude Sonnet 5 and Opus 5 with higher limits, Claude Code, Projects for saved context, and the longest practical context window for document-heavy work. Claude Max starts at $100 if Pro caps you. For explanations, drafts, research summaries and coding questions outside an IDE, DeepSeek's free tier covers much of what the paid plans do. The $20 pays for a feature, not for better answers to typical text prompts.
API prices: where the real gap is
One workload, 10M input and 2M output tokens a month, priced on each API. The DeepSeek rows use peak-hour rates: DeepSeek V4.1 Flash ($0.30/$1.20): $3.00 + $2.40 = $5.40 DeepSeek V4 Pro ($1.32/$3.96): $13.20 + $7.92 = $21.12 Claude Sonnet 5 ($2/$10): $20 + $20 = $40 GPT-5.4 ($2.50/$15): $25 + $30 = $55 GPT-5.5 ($5/$30): $50 + $60 = $110 That makes V4.1 Flash about 7 times cheaper than Sonnet 5 ($40 ÷ $5.40) and 20 times cheaper than GPT-5.5 ($110 ÷ $5.40). Off-peak, DeepSeek bills half, so the Flash month drops to $2.70. Peak is 01:00-04:00 and 06:00-10:00 UTC, Monday to Friday; every other hour, weekends and Chinese public holidays are off-peak. Prompt caching (cached reads roughly 10% of list at OpenAI and Anthropic) and batch pricing shrink the dollar figures but not the order. The table can't tell you whether Flash is good enough for your task. Only a test on your own prompts can.
Cheapest Option
DeepSeek V4.1 Flash is $0.30 per 1M input tokens and $1.20 per 1M output at peak, the cheapest API on this page. It replaced V4 Flash; the API name is now deepseek-flash. V4 Pro at $1.32/$3.96 is the step up if Flash misses your quality bar. Test both on your own prompts before you move production traffic.
Our Recommendation
Use DeepSeek's free chat as your default for a month. Buy ChatGPT Plus only if you reached for voice or image generation in that month, or Claude Pro if you coded with Claude Code; buy one, not both. Before you paste work or client documents into any free chat app, read its privacy terms. For a product, run your real prompts through DeepSeek V4.1 Flash, then V4 Pro. Pay Claude Sonnet 5 or GPT-5.5 rates only for the calls where they fall short.
If you're picking today: start with the cheaper viable option, then validate your monthly usage in the calculator before committing.
Tradeoff Matrix
| Use case | Cheapest | Best value | Most reliable | Easiest to start |
|---|---|---|---|---|
| Free-tier personal use, text-only | DeepSeek (free) | DeepSeek (free) | ChatGPT Plus | DeepSeek web chat |
| Voice/image-heavy personal use | None: DeepSeek has no native voice or image generation | ChatGPT Plus ($20/mo) | ChatGPT Plus | ChatGPT app |
| Long-document/coding personal use | None: DeepSeek's free chat has no coding agent | Claude Pro ($20/mo, Claude Code included) | Claude Pro | Claude Code |
| Building a product (API) | DeepSeek V4.1 Flash ($0.30/$1.20 peak) | DeepSeek V4.1 Flash or V4 Pro | Claude Sonnet 5 / GPT-5.5 | OpenAI API (largest ecosystem) |
Break-Even Analysis
free Vs Paid
For text-only use there is no break-even: DeepSeek costs $0, so a $20 plan only pays off through a feature DeepSeek lacks. A simple test: if you didn't use voice, image generation, Claude Code or a long-document Project last month, try a month on DeepSeek before you renew.
api Threshold
On the API, your chat plan is irrelevant; you pay per token. The choice is DeepSeek against GPT-5.5 (or GPT-5.4) against Claude Sonnet 5 on your own workload. At the 10M-input, 2M-output month above, moving from Sonnet 5 to V4.1 Flash saves about $35 a month ($40 − $5.40), and from GPT-5.5 about $105 ($110 − $5.40). Our DeepSeek vs GPT-5.5 and DeepSeek vs Claude pages take the math further.
Recommendation by Buyer Type
Casual personal user
DeepSeek free tier
Costs nothing; the only cap is peak-hour throttling, and it covers everyday text tasks
Wants voice, image generation or GPTs
ChatGPT Plus ($20/mo)
DeepSeek's app doesn't match GPT-5.5's voice mode or image generation
Codes daily and wants an agent
Claude Pro ($20/mo) for Claude Code
A terminal coding agent on Sonnet 5 comes with the plan; DeepSeek's free chat has none
Building an API-powered product
Skip the consumer plans; test DeepSeek V4.1 Flash and V4 Pro against GPT-5.4, GPT-5.5 and Claude Sonnet 5 on your workload
The sample month costs $5.40 on V4.1 Flash and $110 on GPT-5.5, a gap worth five $20 plans
Who is overpaying for ChatGPT Plus?
- You pay $20 for ChatGPT Plus or Claude Pro and use it only for text questions and drafts that DeepSeek's free tier would handle
- You pay for both plans for the same personal use, instead of keeping the one whose feature (voice and images, or Claude Code) you use
- You build on GPT-5.5 or Claude Sonnet 5 API rates without first testing whether DeepSeek V4.1 Flash or V4 Pro clears your quality bar
- You assumed your ChatGPT Plus or Claude Pro plan includes API credits. It doesn't; API use bills separately
Bottom line
Pay $0 for text. Pay $20 only for a feature you will use this month: voice and images on ChatGPT, Claude Code on Claude. On the API, start with V4.1 Flash at $0.30/$1.20 (peak) and move up a price tier only when your own tests show you need to.
What users report
Users praise DeepSeek's low per-token price, but some say peak-hour pricing, fix-up tokens, and China data handling erode the savings.
Paraphrased from public posts, with the date and a link to each original. These are other people's accounts, not our tests, and we have not verified each one.
One user with both DeepSeek's API ("hooray off-peak pricing") and a $100 Claude subscription said they mostly use DeepSeek Flash because it is fast and nearly always correct.
Value · DeepSeek V4.1 Flash vs Claude Max · Hacker News, Oct 2026
One user said they use Claude Opus at work and DeepSeek V4.1 Flash for personal projects; for their kind of work the two are "basically interchangeable" and DeepSeek costs 3% of what Claude does.
Value · DeepSeek V4.1 Flash vs Claude Opus · Hacker News, Nov 2026
After months on DeepSeek 4.1 Flash, one user tried Claude Sonnet 5.5 and found it right first time. They said DeepSeek's extra fix-up tokens made it as expensive or more, before counting their own time.
Quality · DeepSeek V4.1 Flash vs Claude Sonnet · Hacker News, Oct 2026
A user in Singapore said DeepSeek became less useful after its price increase, because their working hours overlap China's peak-pricing hours, though a stronger new Flash model made it somewhat cheaper in effect.
Billing · DeepSeek API · Hacker News, Sep 2026
One user said OpenCode Go suddenly blocked DeepSeek unless they accepted their data going to China, which weakened their trust; they also found DeepSeek "so much more expensive" than before.
Value · DeepSeek via OpenCode Go · Hacker News, Sep 2026
One user preferred a European host for DeepSeek Flash 4.1, paying Mistral $0.03/$0.007/$1.20 versus DeepSeek's own peak $0.03/$0.006/$1.20 and off-peak $0.015/$0.003/$0.60 (their figures).
Billing · DeepSeek V4.1 Flash (Mistral-hosted) · Hacker News, Oct 2026
Sources
Prices on this page are checked against these vendor pricing pages and catalogue entries.
Frequently Asked Questions
Is DeepSeek really free?
The web chat is, at $0 with no message cap; the only limit is fair-use throttling at peak hours. The API is paid: DeepSeek V4.1 Flash costs $0.30 per 1M input tokens and $1.20 per 1M output at peak, and V4 Pro $1.32/$3.96. Both cost half off-peak.
Does ChatGPT Plus or Claude Pro come with API credits?
No. Both are chat plans. API use is billed per token on a separate account at the rates on this page, even if you already pay $20 for the app.
Is DeepSeek's API still cheaper than GPT-5.5 with prompt caching?
Yes. Cached reads at OpenAI cost about 10% of list, which cuts GPT-5.5's input from $5 to about $0.50 per 1M. That is still above V4.1 Flash's $0.30 peak input rate. Output stays at $30 per 1M against $1.20, so the ranking holds.
Should I pay for both ChatGPT Plus and Claude Pro?
Rarely. That is $40 a month. Keep the one whose feature you use: voice and images on ChatGPT, or Claude Code and long documents on Claude. Use DeepSeek's free chat or the other vendor's free tier for the rest.
Editorial context
Who is this for?
Developers, startups, and teams who want to reduce their AI API or subscription costs without sacrificing quality.
When NOT to use this
Users who need real-time data, image generation, or proprietary enterprise integrations may need more specialised tools.
Pricing insights
AI pricing varies widely — some models charge per token while others use flat subscriptions. Token-based APIs are usually cheaper for moderate usage, while subscriptions suit power users with high and consistent volume.
Alternatives to consider
Consider DeepSeek V4.1 Flash for cost-effective coding and writing, Gemini 3.8 Flash for fast tasks, or Claude Haiku 4.5 for lightweight structured work. Use the calculator to compare your specific usage.
Final verdict
The cheapest AI tool is the one that fits your exact workload. Use the cost calculator and decision engine on this site to find your optimal stack — most users can cut AI spend by 50% or more.
Related
Free courses · no sign-up
Still deciding? Learn the basics first, then come back to the prices.
Pricing based on publicly available rates. Check current provider pricing before subscribing. Some links may be affiliate links — see our affiliate disclosure.