01What happened

The story, straight

Z ai's GLM-5.2 is now the leading open-weight model on the Artificial Analysis Intelligence Index, scoring 51 — 11 points higher than its predecessor GLM-5.1. The model, which has 744B total parameters with 40B active, outperforms MiniMax-M3 (44), DeepSeek V4 Pro (44), and Kimi K2.6 (43) on the v4.1 index. GLM-5.2 is priced at $1.4/$4.4/$0.26 per 1M input/output/cache hit tokens on its first-party API, matching GLM-5.1 pricing. The biggest gains came in scientific reasoning: +16 points on CritPt (to 21%) and +12 points on HLE (to 40%).

z ai just dropped glm-5.2 and it's now the top open-weight model on Artificial Analysis, scoring 51 on the intelligence index — 11 points above glm-5.1. same architecture size (744b total, 40b active), just way smarter. it beats minimax-m3 (44), deepseek v4 pro (44), and kimi k2.6 (43). api pricing stays flat at $1.4/$4.4/$0.26 per 1m tokens. the real jump is in scientific reasoning: +16 points on critpt, +12 on hle.

02Spread timeline

Where it actually started

Jun 17, 2026Origin
Hacker News post surfaces GLM-5.2 Artificial Analysis article.hn post surfaces the glm-5.2 benchmark write-up
source
Jun 17, 2026
Artificial Analysis publishes Intelligence Index update naming GLM-5.2 as top open-weight model.artificial analysis officially names glm-5.2 the leading open weights model
source

03Source receipts

Every claim, linked

Artificial AnalysisPrimary write-up: GLM-5.2 scores 51 on Intelligence Index v4.1, with detailed benchmark breakdowns including scientific reasoning gains.
primaryhackernewssignal
Hacker NewsHacker News discussion thread linking to the Artificial Analysis article.
supportingmastodonsignal

04What's solid, what isn't

What's solid and what isn't

Confirmed
  • GLM-5.2 scores 51 on the Artificial Analysis Intelligence Index v4.1.
  • GLM-5.2 has 744B total parameters with 40B active, same architecture size as GLM-5.1.
  • GLM-5.2 outperforms MiniMax-M3 (44), DeepSeek V4 Pro (44), and Kimi K2.6 (43).
  • API pricing is $1.4/$4.4/$0.26 per 1M input/output/cache hit tokens.
Disputed
  • The full list of benchmark improvements beyond scientific reasoning (AA-LCR, tau3 banking data partially cut off in source).

05Why it matters

The editorial take

Open-weight AI models have been closing the gap with proprietary systems, and GLM-5.2's jump suggests Z ai found significant efficiency gains without scaling up parameter count. The pricing stability at top-benchmark performance is notable for developers building on open weights — it means better capability without higher API costs. Scientific reasoning improvements could signal broader gains in research-adjacent applications.

open-weight models keep getting closer to the closed ones and glm-5.2 is the latest.