Artificial Analysis Progress Index: US-China Frontier Gap Narrows to ~3.3 Months, Chinese Open-Weight Models Lead

Artificial Analysis launched Progress Index v1, a unified scale measuring LLM capability progress since November 2022, revealing three key findings: over four years the frontier score rose from GPT-3.5 Turbo's 125 to 325; the US-China frontier gap is only about 3.3 months; and open-weight models keep closing in on closed ones, with Chinese open-source models leading since September 2025. The index provides a quantitative benchmark for watching the AI competitive landscape.

Confirmed

• Artificial Analysis introduced the Progress Index, using item response theory (IRT) to estimate general model capability, released alongside Intelligence Index v5 (m3, m6)
• Scale examples: GPT-3.5 Turbo from November 2022 scores 125; the first model to reach 300 was GPT-5.3 Codex (priced at $14 as of February 2026); DeepSeek V4.1 Flash 1017 scores 324 (m3, m4, m7)
• The US-China frontier gap of roughly 3.3 months is an estimate by scaling01 based on Progress Index data (m1)
• Open-source catch-up pace: since January 2025, each time an open-weight model set a new high score, it lagged the first proprietary model to reach that level by 2-8 months (m2)
• Battle for open-source top spot: between July 2023 and September 2025, Chinese and US models traded the lead seven times, most recently with DeepSeek pulling ahead; since then Chinese open-source models have dominated the leaderboard (m5, m7)
• Price collapse: output token prices within the same capability tier dropped by orders of magnitude within months; the 300-score tier fell from $14 to $0.28 (about 50x) (m2, m4)

Why it matters

• The Progress Index offers a unified cross-vendor, cross-time capability scale, making questions like "pace of progress" and "catch-up gap"—previously only discussed qualitatively—quantitatively comparable for the first time
• A US-China frontier gap of 3.3 months and Chinese open-source models on top directly challenge the default narrative of significant US leadership, with implications for policy and investment judgments
• A 50x price drop for the same capability level within months means the marginal cost of AI capability is rapidly approaching zero, which will profoundly reshape application-layer economics

2026-10-29 ~ 2026-10-30 · 7 related posts

• US-China frontier AI gap narrows to just ~3.3 months per AA Progress Index — scaling01 · 2026-10-29
• [source] Progress Index: Frontier AI Capability Climbed From 125 to 325 in Four Years — ArtificialAnlys · 2026-10-30
• [source] Same-Capability Output Prices Fell 50x in Months: $14 to $0.28 Per Million Tokens — ArtificialAnlys · 2026-10-30
• Open weights trail proprietary LLMs by 2-8 months as same-capability prices fall 50x — ArtificialAnlys · 2026-10-30
• [source] China has led open-weights LLMs since Sept 2025 as DeepSeek tops US best by 17 points — ArtificialAnlys · 2026-10-30
• China has led open-weights AI since Sept 2025 as DeepSeek V4.1 Flash 1017 scores 324 — ArtificialAnlys · 2026-10-30
• Artificial Analysis launches Progress Index, a single capability scale for every LLM since 2022 — scaling01 · 2026-10-30