Open-source AI · October 2026
China has made the best open-source AI model every day for two years
Six Chinese labs have passed the lead between them 15 times, most recently to Kimi’s K3.1 Pro. The best American open model is about three months behind.
On 19 Sep 2024, Alibaba released Qwen2.5 Instruct 72B, and it passed Meta’s Llama 3.1 on the Artificial Analysis Intelligence Index, a blend of hard tests in reasoning, knowledge, coding and agent work. Since then, the best open-source AI model, the kind anyone can download and run, has come from a Chinese lab every single day: 765 days in a row.
In 2023, open source meant Llama. Press play to watch the chart turn over, all the way to 23 Oct, when Kimi released K3.1 Pro with a score of 55.1.
Kimi, 55.1
Open-weights models only. The best score among open models each lab had released by that date, since Llama 2 (18 Jul 2023), on the Artificial Analysis Intelligence Index (current version, highest effort setting). The axis rescales as scores rise; the top 10 labs are shown.
Six labs, one country
The lead doesn’t belong to one company. Alibaba, DeepSeek, Kimi, Z.ai, MiniMax and Xiaomi have each held it. DeepSeek had it longest, 307 days in total, most of it with R1, the reasoning model that briefly knocked Nvidia’s share price in January 2025. Z.ai, the lab behind GLM, is second with 198. Since September 2025 there have been 14 different No. 1s, about one every 30 days, and none lasted more than 72 days. The latest, K3.1 Pro, took it from Xiaomi’s MiMo-V2.6-Pro after 32 days.
Hover or tap a stretch to see the model.
The open-weights model with the highest Artificial Analysis Intelligence Index score among those released by each date, 1 Sep 2024 to 23 Oct 2026.
America is three months behind
On 15 Oct, Meta released Muse Spark Open, its first open model since Llama 4 in April 2025. It scores 38.9. A Chinese lab first beat that 91 days earlier, with Kimi K3. That makes it the best open model from anywhere outside China, ahead of South Korea’s Motif 3 (33.6).
It had been worse. In June 2026, the best non-Chinese open model was Google’s Gemma 4 26B A4B, a model small enough for a laptop, and it matched what China had released 246 days, or about eight months, earlier. OpenAI’s gpt-oss-120b, in August 2025, never topped this index.
Artificial Analysis Intelligence Index (current version) of the best open-weights model released by each date. Blue shading is China’s lead; orange, the 18 days in September 2024 when Meta’s Llama 3.1 Instruct 405B led. Hover or tap for the models.
Measured in time, the rest of the world’s best open model was a median of 109 days behind China’s in 2025 and 145.5 days this year. Meta’s release brings it back to about three months.
Days since a Chinese lab first released an open model scoring at least as high as the best open model from anywhere else, weekly. Below zero: a non-Chinese model led, and the number is days until a Chinese lab passed it.
The other tests agree
One index can be gamed, so we checked Epoch AI, a research group that runs its own tests on new models and collects results from others. It has 24 benchmarks with at least 10 open models from 4 or more labs since September 2024. On 8 of them, a Chinese model has had the top open score on every day Epoch has open results for, which ranges from 5 to 20 months. On 11, the exceptions add up to two days or fewer.
The exceptions come from two stretches. In late 2024, Meta’s Llama, Mistral Large and Microsoft’s Phi-4 still led some tests. From August 2025, OpenAI’s gpt-oss-120b led four, in maths, agent work and machine-learning tasks, for four to eight months. Since it was passed, the only outsider to lead one for more than a few days is Google’s Gemma 4, for up to three weeks in April.
The best model you can own
Once the weights are out, anyone can run an open model or serve it, which keeps prices down. For two years the best one has come from China, and the newest is also the closest an open model has come to the best closed one: 31 days behind (more on Kimi K3.1 Pro).
Method
Intelligence scores are the Artificial Analysis Intelligence Index (current version), taking each model at its highest effort setting, by release date, as of 23 Oct 2026. Only models Artificial Analysis marks as open weights count. Our snapshot is from 18 October; we added Kimi K3.1 Pro, released on 23 Oct 2026 with an index score of 55.1, and count it as open because Kimi says it will publish the weights by 4 Nov 2026. Other models released after 18 October aren’t counted. A model counts from its release date. We count from September 2024 because before that every model scores between 5 and 9 on the current index, too close to rank. The race starts at Llama 2 on 18 Jul 2023 and shows each lab’s best open model on each date. Labs are placed by where they are based; a fine-tune counts for the lab that released it, wherever its base model came from.
Benchmarks are from Epoch AI’s Benchmarking Hub (CC BY 4.0), fetched 21 Oct 2026: every benchmark in its metadata with a score column, not superseded by a newer version, with at least 10 open-weights models (as Epoch labels them) released since September 2024 from at least 4 labs, at least one of them Chinese and one not. A model holds the top open score from its release date until a later open model scores higher. Each benchmark is counted to the release date of its newest open model, so one Epoch stopped running isn’t counted after that. Country is Epoch’s. The data and code are in our repository (data/epoch-history, scripts/epoch-open-leads.ts).