01.AI pricing

About

01.AI (零一万物 / Lingyiwanwu) is a Chinese foundation-model company founded in March 2023 by Kai-Fu Lee — the former head of Google China and founder of Sinovation Ventures. It builds the Yi model family: open-weight models (Yi-34B, Yi-1.5, Yi-VL, Yi-Coder) released free under Apache 2.0, alongside proprietary flagship models (Yi-Lightning, Yi-Large) served through a per-token API. It also ships consumer apps — Wanzhi (万知) and the international PopAI — and, increasingly, sells tailored enterprise solutions to industry verticals.

01.AI reached unicorn status (over $1B valuation) within eight months of launching, with backers including Alibaba and Xiaomi. Kai-Fu Lee’s cost-efficiency thesis is central to its pricing identity: the company has claimed it trained a top-tier model on about 2,000 GPUs for roughly $3 million, versus an estimated $80–100 million for comparable frontier runs — a gap it converts directly into cheap token rates.

The strategic story since 2024 is a deliberate retreat from the frontier. After Yi-Large (May 2024) and Yi-Lightning (October 2024) made 01.AI a price-war protagonist, the company scaled back independent frontier pre-training in early 2025, formed a joint “industrial large model laboratory” with Alibaba Cloud (Alibaba trains the giant models; 01.AI builds smaller, cost-efficient ones), and reoriented toward sales-led enterprise vertical solutions in finance, energy, and gaming. That pivot — and the 2024 shutdown of its international API platform — is why a foundation-model lab that once published headline token prices now presents as a largely gated, enterprise-first operation. It sits in the same sovereign-lab cluster as Mistral AI, but where Mistral kept its public token card front-and-center, 01.AI moved inference pricing behind a China-first platform and a solutions motion.

For the latest on 01.AI’s models and platform, visit 01.AI / Lingyiwanwu.

Pricing summary : free open weights, a cheap per-token API, now enterprise-led

01.AI’s pricing has three layers, and they have shifted in emphasis over time:

• Open weights (free). The Yi open-weight family — Yi-34B, Yi-1.5 (6B/9B/34B), Yi-VL multimodal, and Yi-Coder — is published on HuggingFace under Apache 2.0. You download and self-host for free; your only cost is your own compute. This was 01.AI’s first go-to-market (Yi-34B, November 2023) and remains the most transparent layer.
• Closed Yi API (per token). Proprietary flagship models bill per million tokens. Yi-Lightning was cut to about ¥0.99/1M tokens (~$0.14) in October 2024; Yi-Large launched at ¥20/1M (~$2.7) in May 2024, with a tiered ladder (Yi-Large-Turbo, Yi-Medium, Yi-Medium-200K, Yi-Vision, Yi-Spark). After the international platform closed in 2024, granular live rates are gated to the Chinese platform.lingyiwanwu.com.
• Enterprise vertical solutions (sales-led). Since 2025 the primary motion is custom, quoted industry solutions — finance, energy, gaming — built with Alibaba Cloud. There is no public price; engagements are scoped and priced through sales.

What makes this different: 01.AI is a foundation-model lab that moved the wrong way on transparency — from public open weights and headline token prices toward a gated, solution-selling posture — because its founder concluded that racing on frontier pre-training was uneconomic. The cheap token rate (¥0.99/1M) is real and historic, but it is no longer the company’s front door.

Because the current granular RMB card is not publicly renderable, this page sets price_transparency: gated and labels reconstructed rates with their source and date rather than implying a live public sheet.

Pricing by product

Open-weight Yi models (free to self-host)

Third-party hosts (e.g. CloudPrice) price hosted Yi-34B around $0.90 in / $0.90 out per 1M — but the weights themselves are free.

Closed Yi API — proprietary models (per million tokens)

Granular current RMB rates for the non-Lightning tiers are not publicly renderable — gated to the Chinese platform after the August 2024 international shutdown.

Enterprise & consumer surfaces

Sales motions across products: self-serve for open weights and the per-token API; sales-led for enterprise vertical solutions, which are now the company’s primary revenue motion.

Hidden costs : What 01.AI users actually pay

01.AI’s headline numbers look almost free — open weights at $0 and Yi-Lightning at ~$0.14/1M — but the real cost depends heavily on which layer you use. Two archetypes show how the total assembles.

Archetype 1 — a developer self-hosting Yi-34B. The weights are free, but you pay for GPUs. Assume one A100-class instance running roughly 720 hours/month at a typical cloud rate.

The lesson: “free open weights” shifts cost from a per-token line to your own infrastructure. For low volume, the closed API is far cheaper; self-hosting wins only at scale or where data residency demands it.

Archetype 2 — a developer calling Yi-Lightning over the API. At ¥0.99/1M (~$0.14), even heavy usage stays cheap — but the friction is access, not price.

Here the surprise is not the bill — it’s the gate. After the international platform closed in 2024, the cheapest path runs through the Chinese platform.lingyiwanwu.com, which is region-oriented and does not surface an open English RMB card. The “hidden cost” is the onboarding and region friction, not token spend.

Want to estimate your own 01.AI bill? Use the 01.AI pricing calculator to model token volume against the Yi-Lightning and Yi-Large rates, or weigh self-hosting compute versus the per-token API.

Pricing evolution : 01.AI pricing history and changes

01.AI’s pricing arc is unusually sharp: open weights first, then a price-war sprint to the cheapest flagship token rate in China, then a retreat behind a gated platform and a sales-led enterprise pivot. The milestones below are reconstructed from official launch announcements and contemporaneous press.

Cadence

Tracked range: 2023 Q4–2026 Q2. Quarters not listed had no publicly announced price or SKU change. RMB rates labeled with approximate USD; granular current card is gated.

Notable changes

• 2023-11 — Yi-34B open weights released free under Apache 2.0; ranked #1 on HuggingFace’s pretrained base board (open weights, not a paid API, as the first go-to-market).
• 2024-05 — Yi-Large closed API launches at ¥20/1M (~$2.7), under a third of GPT-4 Turbo; Kai-Fu Lee declares “cash burning is no longer a winning strategy.”
• 2024-08 — International platform.01.ai suspends API services and routes users to the Chinese platform.lingyiwanwu.com — the first narrowing of the public, English-card API surface.
• 2024-10 — Yi-Lightning cut to ¥0.99/1M (~$0.14), ~40% faster than the prior flagship; LMSYS Chatbot Arena top-6 globally / #1 in China. The signature 01.AI beat in the 2024 price war.
• 2025-01 — Reports that 01.AI is selling its pre-training team to Alibaba; Kai-Fu Lee denies a sale, reframing the Alibaba relationship as a partnership.
• 2025-03 — Joint “industrial large model laboratory” with Alibaba Cloud; frontier pre-training scaled back; pivot to sales-led enterprise vertical solutions (finance, energy, gaming).

The price-war sprint, then the retreat — in detail

The two-step from ¥20/1M (Yi-Large, May 2024) to ¥0.99/1M (Yi-Lightning, October 2024) is the entire story of China’s 2024 LLM price war compressed into one lab. Yi-Large arrived already cheap relative to GPT-4 Turbo; five months later Yi-Lightning was roughly 20× cheaper still, matching peers like Alibaba Qwen, ByteDance Doubao, DeepSeek, and Zhipu in a race toward ~¥1/1M and below. But unlike Mistral — which doubled down on a public token card — 01.AI concluded the frontier race was uneconomic and retreated from public token pricing entirely: the international platform closed, the China platform stopped surfacing an open card, and the company reframed itself around quoted enterprise solutions. The pricing lesson is that winning a price war does not guarantee you stay in the business of selling priced tokens.

What’s unique : 01.AI’s distinctive pricing mechanics

1. The cheapest flagship token in the 2024 China price war. Yi-Lightning at ¥0.99/1M (~$0.14) wasn’t just cheap — it was a top-6 LMSYS Arena model priced below GPT-4o-mini. 01.AI deliberately decoupled benchmark rank from price, using a claimed $3M / 2,000-GPU training cost to justify a rate that closed-frontier labs couldn’t match. The price was the product story.

2. Open weights as the transparent layer, closed API as the priced layer. Like Mistral AI, 01.AI gives the open Yi models away under Apache 2.0 and charges for hosted inference on its closed flagships. The difference is emphasis: the open layer is the most legible thing 01.AI publishes, while its closed API pricing has become less visible over time — the reverse of the usual maturation toward published rates.

3. A foundation lab that priced its way out of frontier pricing. 01.AI’s most distinctive move is strategic: after winning on token price, it concluded that selling priced tokens against giants was a losing game and shifted to sales-led vertical solutions with Alibaba doing the heavy training. The “pricing mechanic” that matters most now isn’t a token rate at all — it’s a quoted enterprise engagement, with the cheap API repositioned as a feeder rather than the business.

Strengths & weaknesses

Billing UX : 01.AI billing controls and transparency

• Open-weight layer — zero billing UX: download from HuggingFace, self-host, and you only ever see your own cloud provider’s bill. The most transparent path 01.AI offers.
• Yi API (China platform) — pay-as-you-go per token on platform.lingyiwanwu.com, which advertises “easy access & flexible pricing.” Region-oriented; an open English RMB card does not render publicly, so prospective international developers can’t self-serve a quote without onboarding.
• International platform — the former platform.01.ai was suspended in 2024, so there is no longer a global English billing surface; users were routed to the Chinese platform.
• Enterprise solutions — no self-serve billing; scoped and invoiced through sales, with deployment and integration bundled into a quoted engagement.
• Currency — pricing is RMB-native (¥); USD figures on this page are approximate conversions at ~¥7.1/$1 and are not official quotes.

Strategic wins : Why 01.AI’s pricing decisions worked

1. Winning the benchmark-per-dollar story

Pricing a top-6 LMSYS Arena model (Yi-Lightning) at ¥0.99/1M let 01.AI own the “frontier quality at commodity price” narrative in China — a clean, headline-able claim that turned a cost-efficiency thesis into marketing. For a period it made 01.AI the reference point for cheap-but-good. See usage-based pricing strategy for why metering inference rather than the model itself scales.

2. Open weights as a distribution flywheel

Releasing Yi-34B and the Yi-1.5 family free under Apache 2.0 — with Yi-34B topping HuggingFace’s base board — bought developer mindshare and credibility that no ad spend could. The open artifact seeded adoption while the closed flagships and enterprise solutions carried revenue. This mirrors the shift away from rigid per-user licensing toward open, adoption-first distribution.

3. Knowing when to stop racing

The most underrated win is the retreat: rather than burn capital chasing frontier pre-training against giants, 01.AI partnered with Alibaba Cloud and redirected to vertical solutions where smaller models win on depth. Choosing not to compete on the most expensive axis — and repricing around solutions — is a discipline most labs learn too late. See choosing the right usage metric for the broader framing.

Areas to improve : Gaps in 01.AI’s pricing approach

1. Restore a public, current token card

The single biggest gap is transparency regression: a lab that once led on headline token prices now hides its current API card behind a region-gated platform. Publishing a live, English-accessible RMB/USD price sheet — even for Yi-Lightning alone — would remove exactly the bill-shock and unpredictability friction that turns evaluators away before they ever spend a token.

2. Give international developers a real on-ramp

Suspending the international platform in 2024 cut off non-China developers from the cheapest path to Yi models. A lightweight global API tier — or a clear, documented route into the China platform — would recover the developer funnel that the open-weight reputation keeps filling. Compare how other AI companies stage their pricing transparency.

3. Expose an enterprise floor or worked example

The enterprise vertical-solutions business is fully sales-gated with no public anchor. A published starting price band or a worked finance/energy deployment example would shorten evaluation for mid-market buyers who can’t justify a sales call but have outgrown self-hosting open weights.

Key takeaways

1. Cheap can be the whole pitch — until it isn’t. Yi-Lightning at ¥0.99/1M made 01.AI the price-war benchmark, but winning on token price did not keep the company in the business of selling priced tokens. A headline rate is a moment, not a moat.
2. Open weights and a priced API are complementary layers. The free Apache-2.0 Yi models seed adoption and credibility; the closed flagships and solutions carry revenue. Give away the artifact, charge for delivery — but keep the priced layer visible.
3. Transparency can regress. 01.AI moved from public open weights and headline prices toward a gated, China-first platform — a reminder that pricing visibility is a choice a company can walk back, often to its own funnel’s cost.
4. Cost structure is a pricing weapon. A claimed $3M / 2,000-GPU training run is what made ¥0.99/1M credible. When your unit economics are genuinely lower, aggressive pricing is defensible rather than suicidal.
5. Knowing which race to quit is a pricing decision. Capping frontier training near Yi-Lightning scale and repricing around vertical solutions let 01.AI compete on depth instead of capital — a deliberate retreat from the most expensive axis.

UBP implications

1. A price-war winner can still exit usage pricing. 01.AI shows that achieving the lowest token rate doesn’t lock you into per-token monetization — macro economics (frontier training cost) can push even a usage leader toward solution selling. UBP strategists should treat a cheap meter as a tactic, not a permanent business model.
2. Open weights change the value question from “what” to “where” and “who runs it.” When the model is free to download, the priced dimension becomes hosted inference, region access, and integration — not the model. UBP design has to follow the cost driver customers can’t trivially replicate.
3. Transparency is itself a value metric. 01.AI’s retreat behind a gated platform raised buyer friction even as its rates stayed cheap. For UBP practitioners, a visible, self-serve price card is part of the product — losing it can cost more adoption than a higher number would.

Sources

• 01.AI / Lingyiwanwu — company (EN) (accessed 2026-06-11)
• 01.AI Yi developer platform (China) (accessed 2026-06-11)
• 01.AI on HuggingFace — open weights (accessed 2026-06-11)
• 01.AI on Wikipedia (accessed 2026-06-11)
• Kai-Fu Lee sets the record straight on 01.AI’s pivot — KrASIA (accessed 2026-06-11)
• Alibaba ties up with Lee Kai-fu’s unicorn — SCMP (accessed 2026-06-11)
• Yi-Lightning Technical Report — arXiv 2412.01253 (accessed 2026-06-11)
• Browse the pricing blueprint corpus

Bottom line

01.AI is a Chinese foundation-model lab that priced its way to the front of the 2024 LLM price war — Yi-Lightning at ¥0.99/1M (~$0.14) — and then deliberately stepped back. Its open-weight Yi models stay free on HuggingFace under Apache 2.0, its closed Yi API is now gated to a China-first platform, and its primary motion is sales-led enterprise vertical solutions built with Alibaba Cloud. The cheap token rate is real and historic, but it is no longer the company’s front door — a cautionary case in how a usage-pricing leader can retreat from public, self-serve pricing even after winning on price.

Want to compare 01.AI against other foundation-model labs? See Mistral AI, or browse the full pricing blueprint.