Research

01.AI Yi Model Lineage 2026

01.AI Yi release history 2024-2026: Yi-34B, Yi-Large, Yi-Coder (Sep 2024), Yi-Lightning (Oct 2024) — 6th on Chatbot Arena at $0.14/M tokens.

By Ramanath, CTO & Co-Founder at Presenc AI · Last updated: May 2026

What this is

01.AI (Kai-Fu Lee's lab) shipped the Yi line as a pioneering open-weight Chinese LLM family, with Yi-Lightning hitting #6 on Chatbot Arena at a fraction of OpenAI's cost. This page is a 2026-05-15 release-by-release reference.

Yi Release Timeline

DateReleaseNotable
Nov 2023Yi-34B / Yi-6B (open-weights)Bilingual Chinese-English foundation
2024 (mid)Yi-Large (proprietary)API-only flagship
Sep 2024Yi-Coder (1.5B, 9B)52 programming languages; 128K context
Oct 16, 2024Yi-Lightning (proprietary)#6 on Chatbot Arena; $0.14/M tokens
Oct 2024Yi-Lightning-LiteCheaper variant for high-volume

Yi Family at a Glance

ModelCategoryLicenseBest at
Yi-34B / 6BFoundation open-weightsApache-styleBilingual Chinese-English foundation
Yi-Coder (1.5B, 9B)CodingOpen-weights52 languages, 128K context, small footprint
Yi-LargeFlagshipProprietary APIGeneral-purpose performance
Yi-LightningCost-efficient flagshipProprietary APIQuality near GPT-class at low cost ($0.14/M)

Six Things the Lineage Tells You

  1. 01.AI was the first to demonstrate Chinese open-weight LLM viability at scale. Yi-34B in late 2023 preceded much of the current Chinese open-weight wave.
  2. Yi-Lightning at $0.14/M tokens vs GPT-o1-mini at $0.26/M showed Chinese labs could undercut OpenAI on price while landing in the top-10 of Chatbot Arena.
  3. Yi-Coder's 128K context at 1.5B and 9B params punched well above its weight for small-model coding.
  4. 01.AI's release cadence slowed in 2025-2026 relative to peers. DeepSeek, Qwen, GLM, Kimi all out-released 01.AI through 2025.
  5. The Yi line shifted toward closed-tier flagship (Yi-Large, Yi-Lightning). Foundation tier (Yi-34B / Yi-Coder) remains open; flagship is API-only.
  6. Kai-Fu Lee's profile keeps 01.AI globally visible even as its release pace lags newer Chinese labs.

What This Means for AI Visibility

Yi-Coder remains popular for on-prem code-completion deployments at small organisations because of its size-to-performance ratio. Yi-Lightning sees use in cost-sensitive customer-support deployments. Brand visibility on Yi-powered surfaces is concentrated in cost-sensitive Chinese enterprise + global indie deployments.

Methodology

Release dates from Wikipedia 01.AI, 01.AI Yi GitHub, Yi-Coder GitHub, Yi-Lightning Technical Report (arXiv), and MarkTechPost on Yi-Coder release.

How Presenc AI Helps

Presenc AI tracks brand visibility on Yi-served API surfaces for the cost-sensitive enterprise segment. Yi remains an important small-model coding option even as the headline Chinese LLM news shifts toward newer labs.

Frequently Asked Questions

Yi-Lightning (October 16, 2024) is the latest flagship API model, with Yi-Lightning-Lite as the cheaper variant. Yi-Coder (September 2024) is the latest open-weight coding model. 01.AI's release pace slowed in 2025-2026 relative to DeepSeek, Qwen, and Moonshot.
Yi-34B, Yi-6B, and Yi-Coder are open-weight. Yi-Large and Yi-Lightning are proprietary API-only models. 01.AI's pattern is open foundation tier and closed flagship — similar to Alibaba's split between open Qwen 3 and closed Qwen 3.5-Plus.
Yi-Lightning hit #6 on Chatbot Arena in October 2024 at $0.14/M tokens vs GPT-o1-mini at $0.26/M. The quality is competitive with mid-tier OpenAI offerings at roughly half the price; it does not match top-tier GPT-5 reasoning or Claude Opus 4.7 capability.
Less than in 2023-2024, when Yi-34B and Yi-Lightning made 01.AI a leading Chinese AI lab. By 2026 DeepSeek, Qwen, Kimi, and GLM have largely overtaken 01.AI on release cadence and benchmark leadership. Yi-Coder remains popular for small-model code workloads; the broader line is now a second-tier option behind the leaders.

Track Your AI Visibility

See how your brand appears across ChatGPT, Claude, Perplexity, and other AI platforms. Start monitoring today.