# China AI Pricing 2026: DeepSeek, Qwen, Kimi API Costs > Current 2026 API and subscription prices for DeepSeek, Qwen, Kimi, GLM, and MiniMax, compared per million tokens and against US models. Source: https://capitalandcompute.net/blog/china-ai-pricing/ Published: 2026-07-18 Updated: 2026-10-04 Chinese AI models list API prices well below their US counterparts, and the gap depends entirely on which two models you put side by side. DeepSeek V4.1 Flash starts at $0.15 per million input tokens off-peak. At the other end, [Kimi K3](/blog/kimi-k3-explained/) lists output at $15\. Line up models of the same measured intelligence and the honest gap runs from about 1.6x to 12x on output tokens, not the flat “90 percent cheaper” you see in headlines. Here is what every major Chinese lab charges in 2026, per token and per month, checked against each provider’s own price page. $0.15 DeepSeek V4.1 Flash input per million tokens off-peak, $0.30 at peak 6.3x DeepSeek V4 Pro vs Claude Opus 4.8 output price gap at peak, 12.6x off-peak 5 labs cut token prices 50 to 99 percent across 2026, per press reports ## What Chinese AI models cost per token These are list rates per million tokens, verified on July 18, 2026 against each provider’s own pricing page, with the DeepSeek rows re-verified on October 4, 2026\. DeepSeek now bills by time of day, so its rows show the off-peak rate first and the peak rate second. Input is what you pay to send tokens, output is what the model generates, and the cache column is the discounted rate for input the provider has already seen (the number that decides agent economics, because agents resend the same context every turn): | Model (lab) | Input | Output | Cache read | | ------------------------- | ----------- | ----------- | ------------- | | DeepSeek V4.1 Flash | $0.15–$0.30 | $0.60–$1.20 | $0.003–$0.006 | | MiniMax M3 | $0.30 | $1.20 | $0.06 | | DeepSeek V4 Pro | $0.66–$1.32 | $1.98–$3.96 | $0.022–$0.044 | | Kimi K2.7 Code (Moonshot) | $0.95 | $4.00 | $0.19 | | GLM-5.2 (Zhipu) | $1.40 | $4.40 | $0.26 | | Qwen3.7 Max (Alibaba) | $2.50 | $7.50 | $0.25 | | Kimi K3 (Moonshot) | $3.00 | $15.00 | $0.30 | Correction, October 4, 2026: DeepSeek rows updated This table previously listed DeepSeek V4 Flash at $0.14 / $0.28 and V4 Pro at $0.435 / $0.87\. Both rates are gone. From August 16, 2026 DeepSeek bills by time of day: peak is 01:00 to 04:00 and 06:00 to 10:00 UTC, Monday to Friday, excluding Chinese public holidays, and every other hour is off-peak at half the peak rate. On September 10, 2026 V4 Flash was retired; its model ids are now answered by V4.1 Flash (`deepseek-flash`) and billed at its rate. Both changes are on [DeepSeek’s own pricing page](https://api-docs.deepseek.com/quick%5Fstart/pricing). Thanks to Jingyu of [AI Daily Insights](https://www.aidailyinsights.cn/en/pricing/), which keeps a dated, sourced change log of Chinese model API prices, for flagging the stale rows. Update, August 7, 2026: Qwen3.8-Max lands below Qwen3.7 Max Alibaba announced Qwen3.8-Max at $2.00 input / $6.00 output per million tokens on August 3, undercutting the Qwen3.7 Max row in the table above. Alibaba Cloud’s Model Studio pricing page still does not list the model, so the rate rests on the Qwen team’s own X post, but Artificial Analysis has now measured that rate on a live endpoint and scored the model at [Intelligence Index 58, ninth of 185 models](/blog/qwen-3-8-max-preview/). The catch fits this page’s argument rather than undercutting it: running the same eval suite cost 64 percent more than on Qwen3.7 Max despite the lower token price. Two things get lost when people quote a single “China is X percent cheaper” number. First, the cache rates. DeepSeek bills a V4.1 Flash cache hit at $0.003 per million tokens off-peak ($0.006 at peak), roughly a 98 percent discount on its own input rate, which makes a long system prompt or a repeated codebase almost free to resend. For a coding agent that ships the same 40,000-token context on every turn, that column matters more than the sticker input price. It is the same force pulling the whole market down that we tracked in [the collapse in cost per token over time](/blog/cost-per-token-over-time/). It is also the line DeepSeek raised hardest in August: the V4 Pro cache hit went from $0.0036 to $0.044 at peak, about 12 times the old rate, a bigger jump than the input or output lines. Second, DeepSeek runs two tiers that get flattened into one “DeepSeek” row elsewhere. V4.1 Flash is the budget workhorse at $0.15 input off-peak. V4 Pro is the larger, pricier model at $0.66 off-peak and $1.32 at peak. Both undercut anything comparable from a US lab, but they are not the same product, and quoting the Flash price against a US flagship is the kind of apples-to-oranges move that produces the 90-percent-cheaper headline. ## China vs US: the real price gap The 50x claim is real and misleading at the same time. It is real if you pick the cheapest Chinese model and the most expensive US one. It is misleading as a general statement, because most buying decisions compare models that do roughly the same job. So match them on capability. Using the intelligence composite from our [AI model value leaderboard](/ai-model-leaderboard/), here is each Chinese model against the US model closest to it in measured smarts, by output price: __Capability-matched output price: Chinese models vs US peers__ | Item | US peer | Chinese model | | ---------------------------------- | ------- | ------------- | | DeepSeek V4 Pro vs Claude Opus 4.8 | $25 | $3.96 | | MiniMax M3 vs GPT-5.3 Codex | $14 | $1.2 | | GLM-5.2 vs GPT-5.6 Terra | $12 | $4.4 | | Qwen3.7 Max vs Gemini 3.1 Pro | $12 | $7.5 | | Kimi K3 vs GPT-5.6 Sol | $30 | $15 | Output list price per million tokens, US dollars, July 2026, with the DeepSeek row updated to its peak rate on October 4, 2026\. Each row pairs a Chinese model with the US model nearest to it on the intelligence composite. The gold dot is the Chinese model, the slate dot its US peer; the line is the price gap.Source: [Provider price pages and the value leaderboard, verified July 2026](/ai-model-leaderboard/) Read the pairs and the story changes row by row. DeepSeek V4 Pro against Claude Opus 4.8 was the widest gap here at 29x on DeepSeek’s old flat rate. The August repricing cut it to about 6.3x at peak and 12.6x off-peak, which leaves MiniMax M3 against GPT-5.3 Codex, at 11.7x, as the widest peak-rate pair. One caveat keeps the DeepSeek row honest: its score is an Artificial Analysis read while the Opus score is a Price Per Token composite, so the capability match is approximate. But Qwen3.7 Max against Gemini 3.1 Pro, two models a rounding error apart on capability, is a 1.6x gap. GLM-5.2 against a mid-tier GPT is about 2.7x, down from 3.4x after OpenAI cut Terra on July 30\. The rule of thumb: peer-to-peer, Chinese models run three to five times cheaper, and only at the top of the US price stack does the multiple pass 10x. Still decisive. A 3x difference on output tokens rewrites a product’s unit economics. It just is not 50x across the board. Update, July 31, 2026: a US model just entered the Chinese price band On July 30 OpenAI [cut GPT-5.6 Luna by 80 percent](/blog/did-openai-cut-ai-prices-gpt-5-6/), to $0.20 input / $1.20 output. That is the same output rate as MiniMax M3, and below GLM-5.2 at $4.40 and Qwen3.7 Plus at $1.28\. It is the first time in this analysis that a US lab has matched the Chinese budget tier on price rather than conceding it. Note the capability comparison crosses benchmark scales (Luna is scored on Artificial Analysis, the Chinese models on the Price Per Token composite), so treat the like-for-like claim as approximate. The peer-to-peer multiples above still hold in the mid and upper tiers, where Terra, Sol and the Anthropic flagships sit. Plot the whole field and the pattern is a cloud that leans one way: __The value frontier: intelligence vs output price__ | Item | intelligence index (Price Per Token composite, Artificial Analysis where noted) | output price per million tokens, USD (log) | | --------------- | ------------------------------------------------------------------------------- | ------------------------------------------ | | DeepSeek V4 Pro | 44 | $3.96 | | MiniMax M3 | 44.4 | $1.2 | | Qwen3.7 Plus | 39 | $1.28 | | Kimi K2.7 Code | 41.9 | $4 | | GLM-5.2 | 51.1 | $4.4 | | Qwen3.7 Max | 46 | $7.5 | | Kimi K3 | 57 | $15 | | Grok 4.5 | 54 | $6 | | Gemini 3.1 Pro | 46.5 | $12 | | GPT-5.6 Terra | 55 | $12 | | Claude Opus 4.8 | 55.7 | $25 | | GPT-5.6 Sol | 59 | $30 | | Claude Fable 5 | 59.9 | $50 | Intelligence composite against output price (log scale), July 2026\. Chinese models in gold cluster in the low-price band at every capability level; US flagships in slate sit above them.Source: [AI model value leaderboard, benchmark composite from Price Per Token, verified July 2026](/ai-model-leaderboard/) The three highest-intelligence models are still American. Everything below them, at any given capability level, has a Chinese option sitting at a fraction of the price. Why that cloud looks the way it does, the mixture-of-experts efficiency, the cheap hub-zone power and the open-weight strategy, is its own story: we walked through it in [why Chinese AI models are so cheap](/blog/why-are-chinese-ai-models-so-cheap/). This piece is about what you actually pay. ## Subscription and coding-plan pricing Per-token billing is not the only way to buy these models. Every major Chinese lab now sells a flat-rate coding subscription, the same model as Claude Code or a Cursor plan: pay a fixed monthly fee, get a rolling usage window of prompts inside your editor, stop watching the meter. For steady daily use they undercut token billing by a wide margin, which is why they matter to the “china ai pricing” question as much as the API rates do. | Coding plan (lab) | Entry price | Runs | | ---------------------- | ----------- | -------------------------- | | MiniMax Starter | \~$10/mo | MiniMax M2.7 | | Kimi Code (Moderato) | $15/mo | Kimi K2.6 | | GLM Coding Plan (Lite) | $18/mo | GLM-5.2, GLM-5-Turbo, 4.7 | | Qwen Code (Pro) | $50/mo | Qwen3.7-Plus, Kimi, others | Two caveats before you pick one. Alibaba retired its cheaper Qwen Lite tier in March 2026, so $50 a month is now the floor for the Qwen plan, and interestingly that Pro plan bundles rival models (Kimi K2.5, GLM-5, MiniMax) alongside Qwen’s own. GLM’s $18 Lite tier gives you roughly 80 prompts per five-hour window, which is plenty for light use and cramped for a full workday. These plans change often, so we track the live cross-vendor comparison, including the US plans, on the [AI coding plan pricing pillar](/ai-pricing/). ## The 2026 China AI price war, in dates None of these numbers hold still. The through-line of 2026 has been Chinese labs cutting, then cutting again, and US labs answering: 1. April 2026 DeepSeek V4 lands 97 percent below GPT-5.5 DeepSeek prices its V4 model at rock-bottom rates, reported by the South China Morning Post as 97 percent below OpenAI’s comparable flagship, and pairs it with Huawei Ascend integration. 2. May 2026 DeepSeek makes a 75 percent V4-Pro cut permanent What started as a promotion becomes the standing rate. Cached input drops toward a few cents per million tokens. 3. May 2026 DeepSeek reaches 17 percent of Vercel token volume Rest of World reports DeepSeek at 17 percent of tokens on Vercel’s AI platform while collecting about 1 percent of the revenue: adoption running far ahead of monetization. 4. Across 2026 Five labs cut token prices 50 to 99 percent ByteDance, Tencent, MiniMax, Alibaba and Xiaomi all cut, per press tallies, with Xiaomi reported slashing some API rates by 99 percent. 5. July 2026 US labs answer in the mid-tier OpenAI positions GPT-5.6 Terra at near-flagship capability for half the top rate, then on July 30 cuts Terra 20 percent and Luna 80 percent. Luna lands at $0.20/$1.20, inside the Chinese budget band for the first time. 6. August 2026 DeepSeek switches to peak and off-peak billing From August 16, V4 Pro output rises from $0.87 to $1.98 off-peak and $3.96 at peak, with peak set to Beijing office hours. 7. September 2026 DeepSeek V4.1 Flash replaces V4 Flash Launched September 10 at $0.15/$0.60 off-peak and $0.30/$1.20 at peak, below the V4 Flash card it retires. The old V4 Flash model ids now route to it. The direction is mostly one-way, and DeepSeek’s August rise is the exception worth watching. When a competitor sells adequate capability at a fifth of your price, you either match on the tiers where you overlap or you cede them. That is the squeeze we unpacked in [modeled cost per finished coding task](/coding-agent-cost-per-task/), where the premium model still wins on the hardest work while the challenger wins on cost per task. ## How to actually buy Chinese AI models Price is the easy part. Access is where the decision actually lives, and you have three routes. **First-party API.** Call DeepSeek, Alibaba, Moonshot, Zhipu or MiniMax directly and you get the list prices above. It is the cheapest path and the one most exposed on compliance: your prompts route through infrastructure inside China and its legal regime, which plenty of Western companies exclude by policy. **A Western reseller or inference host.** The same open weights are served by US and EU hosts, so you can run DeepSeek or Qwen without sending data to China. You pay a markup over the first-party rate for that, and the [AI inference provider directory](/ai-inference-providers/) lists who carries which model at what price. **Self-hosting the open weights.** This is the lever that has no equivalent on the US side. DeepSeek ships under an MIT-style license, and Qwen, Kimi, GLM and MiniMax all release open weights of their own (Kimi K3’s open weights are due July 27, 2026). Download them, run them on your own or rented GPUs, and the per-token API price stops applying entirely: your cost becomes hardware and electricity. For teams with steady, high volume, that is often cheaper than any API rate, Chinese or American. It is also why the [Hugging Face ecosystem now leans Chinese](/blog/why-are-chinese-ai-models-so-cheap/), and why the whole category is hard to price with a single number. The prices are subsidized, and subsidies lift Several of these rates are priced for market share, not margin. DeepSeek is funded by a hedge fund and does not need API revenue to survive. That is great for buyers today and a real risk on a two-year horizon: a rate set to win share can rise once the share is won. Track the verification date, not the assumption that today’s price holds. Our [model price tracker](/ai-models/) stamps every rate with when it was last checked. If you are weighing the other side of the ledger, the models these prices are pressuring, we keep a running view of the [best frontier models excluding Chinese labs](/best-non-china-ai-models/) and the national picture behind the pricing in the [AI usage by country](/ai-nations/) data. ## Bottom line For the “china ai pricing” question, the honest answer is a range, not a slogan. Peer-to-peer, Chinese models cost roughly three to five times less per output token than the US model doing the same job; at the extreme, the gap still passes 10x, as with MiniMax M3 against GPT-5.3 Codex, or DeepSeek V4 Pro against Claude Opus 4.8 off-peak. Coding subscriptions start around $10 to $18 a month against $20 for the cheapest US plan. And for steady volume, the open weights take the API price to zero and leave only your compute bill. The catch is compliance and the durability of subsidized rates, both of which are judgment calls no price table settles for you. ## Frequently asked questions How much does the DeepSeek API cost? As of October 2026, DeepSeek bills by time of day. V4.1 Flash costs $0.15 per million input tokens and $0.60 output off-peak, and $0.30 and $1.20 at peak. V4 Pro costs $0.66 and $1.98 off-peak, and $1.32 and $3.96 at peak. Peak is 01:00 to 04:00 and 06:00 to 10:00 UTC on weekdays. V4.1 Flash cache hits cost $0.003 off-peak, a roughly 98 percent discount that makes repeated context almost free. What is the cheapest Chinese AI model? On list API rates, DeepSeek V4.1 Flash is the cheapest flagship-class option off-peak, at $0.15 input and $0.60 output per million tokens. At peak it costs $0.30 and $1.20, level with MiniMax M3\. Smaller open-weight tiers from Qwen and GLM go lower still, and self-hosting any open-weight model removes the per-token price entirely. Are Chinese AI models really cheaper than ChatGPT and Claude? Yes, but the gap depends on the match. Compared model-for-model on capability, Chinese models run about three to five times cheaper per output token. On the widest capability-matched pair, the gap passes 10x. The flat 90-percent-cheaper headline only holds at the extremes. Do Chinese AI models have subscription plans? Yes. Every major lab sells a flat-rate coding plan: MiniMax from about $10 a month, Kimi Code from $15, GLM from $18, and Qwen at $50 for its Pro tier. Each gives a rolling window of prompts inside your editor instead of per-token billing. Is it safe to use Chinese AI models? Calling a China-hosted API routes your data through Chinese infrastructure and legal jurisdiction, which many companies exclude by policy. Running the open weights yourself, or through a US or EU inference host, keeps data under your own jurisdiction and is how most Western production deployments use these models. ## Sources * DeepSeek (2026). _Models and Pricing_. DeepSeek API documentation. * Alibaba Cloud (2026). _Model Studio model pricing_. Documentation. * Alibaba Cloud (2026). _Model Studio Coding Plan_. Documentation. * Moonshot AI (2026). _Kimi platform pricing_. Documentation. * Moonshot AI (2026). _Kimi K2.6 pricing_. * Zhipu / Z.ai (2026). _Pricing overview_ and _Coding Plan (DevPack)_. Documentation. * MiniMax (2026). _Pay-as-you-go model pricing_. Platform documentation. * Hugging Face (2026). _State of Open Source on Hugging Face: Spring 2026_. * South China Morning Post (2026). _China DeepSeek prices new V4 AI model 97 percent below OpenAI GPT-5.5_ (as reported). * Rest of World (2026). _Low-cost Chinese AI models like DeepSeek gain traction in the U.S._ (as reported). * AI Weekly (2026). _Five Chinese AI labs cut token prices up to 99 percent_ (as reported).