How far behind is open-source AI?

How far behind is open-source AI?

The open-source Pareto frontier over time: each point is an open-weights model that beat every open model before it — plotted by how many months earlier a proprietary model reached the same level.

The vertical position is the gap to the earliest proprietary model with the same Artificial Analysis<br>Intelligence Index (hover any point to see which one). The gap peaked near 10 months around DeepSeek V3<br>(Dec 2024) and has since tightened to ~2–3.5 months as DeepSeek, Kimi, Z AI and MiniMax began<br>shipping close behind the frontier.

Best open today<br>3.4 mo<br>GLM-5.2 ≈ GPT-5.4

Smallest gap on record<br>2.2 mo<br>DeepSeek V4 Pro ≈ Sonnet 4.6

Widest gap<br>9.8 mo<br>DeepSeek V3 ≈ Claude 3 Opus

Frontier-pushing releases<br>23<br>of 324 open models tracked

open-source frontier — best open-weights model at each date

The open-source frontier's time gap behind the equally-capable proprietary model.

Method & caveats<br>Frontier-pusher. A model is included only if its Intelligence Index exceeded every open-weights<br>model released before it. 23 qualify. (Solar Mini, which the naive running-max flags in early 2024, is excluded: at<br>floor-level scores the retroactive v4.1 index ranks it above contemporaneous stronger open models — Mixtral 8&times;7B,<br>Llama-2-70B, Qwen-72B — so it was not actually the leading open model at release.)

Metric. Artificial Analysis Intelligence Index (v4.1), applied consistently to every model across<br>all dates. It is anchored to today's hard benchmarks (GPQA, HLE, terminal-bench, etc.), so 2023 models score low and<br>their ordering at the floor is noisy.

Gap. For a frontier model released on date D with index S, gap =<br>D &minus; T, where T is the earliest date any proprietary model reached index<br>&ge; S — that proprietary model is shown in each point's tooltip.

Note. This is the gap at a given capability level, not the absolute frontier. GLM-5.2<br>(II 51) matches GPT-5.4 from 3.4 months earlier, while the absolute closed frontier (Claude Fable 5, II 60)<br>sits further ahead in raw capability.

Data snapshot: 17 Jun 2026, scraped from<br>artificialanalysis.ai/models/open-source.<br>Source & code: github.com/yaroslavvb/artificial-analysis-oss-lag.