GLM-5.2 is the new leading open weights model on the Artificial Analysis Intelligence Index

Artificial Analysis · article · 2026-06-17 source ↗ Reader 🖥️

theoretical · expository — A data report: it asserts what is true about GLM-5.2's relative standing on named benchmarks, backed by numbers and charts, rather than arguing the reader should do something. The practical upshot (which model to deploy) is left implicit in the ranking and Pareto-frontier framing.

Structure

Key terms

• Intelligence Index v4.1
  Artificial Analysis's composite benchmark score aggregating multiple evals (including CritPt, HLE, AA-LCR, tau3, SciCode, TerminalBench, GPQA Diamond) into one number; v4.1 is a newly revised version 'featuring upgraded benchmarks and new per-task metrics' per AA's own companion article, shifting toward agentic workloads.
• GDPval-AA v2
  AA's benchmark for real-world, agentic knowledge-work tasks, scored as an Elo rating baselined so that human performance = 1000, judged by a rotating panel of frontier-model judges, with agent turn limits raised from 100 to 250 versus the prior version.
• Pareto frontier (of Intelligence vs Cost per Task)
  The set of models that are not dominated by any other model on both axes simultaneously — i.e., no other model is both as intelligent and cheaper. AA credits GLM-5.2 with the lowest cost per task among models 'at its intelligence level.'
• Active parameters AI Unit Economics
  The subset of a mixture-of-experts model's total parameters (40B of 744B) actually used per token/forward pass — used here as the comparability anchor between GLM-5.1 and GLM-5.2 ('same size... but scores 11 points higher').

Propositions

• explicitGLM-5.2 is the leading open-weights model on the Artificial Analysis Intelligence Index v4.1, scoring 51, ahead of MiniMax-M3 (44), DeepSeek V4 Pro (max, 44), and Kimi K2.6 (43).conf: high
  extendsDeepSeek v4, and the end of the OpenAI/Microsoft AGI clause · p1 — DeepSeek-V4-Pro is the largest open-weights model (1.6T params) yet places behind GLM-5.2 (744B/40B active) on the Intelligence Index — among open-weight models, raw size does not track general-intelligence leaderboard rank.
• explicitGLM-5.2 scores 11 points higher than GLM-5.1 on the Intelligence Index v4.1 despite being the same architecture size (744B total / 40B active parameters).conf: high
• explicitGLM-5.2's gains over GLM-5.1 are concentrated in scientific/technical reasoning evals: CritPt (+16 to 21%), HLE (+12 to 40%), AA-LCR (+9 to 71%), tau3 banking (+15 to 27%), SciCode (+7 to 50%), TerminalBench v2.1 (+16 to 78%), and GPQA Diamond (+3 to 89%).conf: high
• explicitGLM-5.2 leads all open-weights models on GDPval-AA v2 (1524, ahead of MiniMax-M3 at 1418 and DeepSeek V4 Pro max at 1328) and is roughly level with the proprietary GPT-5.5 (xhigh reasoning, 1514).conf: high
• explicitGDPval-AA v2 changed its methodology from the original GDPval-AA: Elo is now baselined to human performance at 1000, a rotating panel of frontier-model judges was introduced, and the agent turn limit was raised from 100 to 250.conf: high
• explicitGLM-5.2 is markedly less token-efficient than its open-weight peers, using 43k output tokens per Intelligence Index task (37k of which is reasoning), versus GLM-5.1 (26k), MiniMax-M3 (24k), and Kimi K2.6 (35k).conf: high
  extendsThe Almedra Token-Efficiency Benchmark, Updated For GLM 5.2 · p3 — Both are independent, near-contemporaneous measurements of GLM-5.2's token/cost behavior — this article's task-level finding (43k output tokens/task, up from GLM-5.1's 26k) corroborates, at the Intelligence Index level, the Almedra benchmark's harness-level finding that GLM-5.2 compressed but did not erase the cost/token spread versus GLM-5.1.
• explicitcontestedDespite its higher token usage, GLM-5.2 sits on the Pareto frontier of Intelligence vs Cost per Task, with the lowest cost per task (~$0.46) among models at its intelligence level, versus GLM-5.1 ($0.25), Kimi K2.6 ($0.31), MiniMax-M3 ($0.18), and DeepSeek V4 Pro max ($0.05).conf: high
  contradictsHow the Escalating AI Wars Benefit You · p9 — This article credits GLM-5.2 with the lowest cost per task at its intelligence level (Pareto frontier), while a later item in the corpus claims GPT-5.6's cheaper Terra/Luna tiers already beat GLM on both performance and cost — a direct tension, though it's not certain both pieces mean the same GLM version/pricing snapshot.
• explicitGLM-5.2 keeps the same parameter count and first-party API pricing as GLM-5.1 while expanding the context window from 200K to 1M tokens.conf: high

Arguments

Narratives

• AA says it "visually inspected GLM-5.2's outputs across a range of GDPval-AA tasks" and attaches "a selection below" — four screenshots of task outputs.
  → Supplies qualitative, illustrative backing for the quantitative GDPval-AA v2 score, but as a curated 'selection' rather than a disclosed, representative or randomized sample.

Questions addressed

• ✅ Which open-weights model currently leads on general intelligence benchmarks?
  GLM-5.2, at Intelligence Index v4.1 score 51.
• ✅ Can an open-weights model match proprietary frontier models on real-world knowledge work?
  Yes, per GDPval-AA v2 — GLM-5.2 (1524) is roughly level with GPT-5.5 xhigh (1514).
• ⭕ Does GLM-5.2's higher token verbosity erode its cost advantage?
  Not directly reconciled — the article reports higher per-task token use and the lowest cost per task at its intelligence level side by side, without explaining how the two are compatible (implicitly: low per-token pricing more than offsets the extra tokens).

Gaps

• illogicalThe article's two headline benchmarks — Intelligence Index v4.1 and GDPval-AA v2 — both received major methodology revisions in the same release cycle as GLM-5.2 itself (v4.1 is described elsewhere by AA as 'a shift toward agentic workloads, featuring upgraded benchmarks and new per-task metrics'; GDPval-AA v2 rebaselines Elo, adds a rotating judge panel, and nearly triples the turn limit). The article never states whether the comparison figures for GLM-5.1, MiniMax-M3, DeepSeek V4 Pro, Kimi K2.6 and GPT-5.5 were rescored under the new methodology or carried over from the old one, so 'GLM-5.2 gained 11 points' and 'GLM-5.2 is now level with GPT-5.5' cannot be cleanly attributed to the model versus the yardstick that measures it.
  Read next: Read against extract-almedra-token-efficiency-benchmark-glm-5-2, which flags an analogous single-run/no-baseline problem for a different GLM-5.2 benchmark; look for AA's own methodology writeup on Intelligence Index v4.1 to see if prior-generation models were re-run under it.
• incompleteNo variance, repetition, or error-bar reporting for any of the headline numbers (score 51 vs 44, GDPval-AA 1524 vs 1514, cost ~$0.46/task) — every comparison is a single point estimate, so it's unclear whether a 1-3 point or $0.01-0.05 gap is a real capability/cost difference or within run-to-run noise. The article also gives no detail on how 'cost per task' is computed (which pricing tier, cached vs uncached tokens, which task set).
  Read next: Compare with the Almedra benchmark's own gap about undocumented statistical procedures, and with the corpus's recurring caution (extract-ai-models-disappear-overnight) that industry benchmarks are useful for orientation, not as a substitute for task-specific evals.
• uninformedThe article does not disclose or discuss that Artificial Analysis is both the designer of the benchmarks (Intelligence Index, GDPval-AA) and the sole publisher of the resulting rankings, with no independent replication cited. It also says nothing about what GLM-5.2's release — an MIT-licensed, commercially usable, frontier-capability open-weights model — means for safety or governance, despite the corpus already flagging this exact silence for GLM-5.1.
  Read next: extract-all-of-ais-new-models-and-tools#g2 raises the identical governance silence for GLM-5.1; extract-almedra-token-efficiency-benchmark-glm-5-2#g3 raises an analogous undisclosed-conflict-of-interest gap for a benchmark publisher whose own product is declared the winner.

Related documents

• The new GPT-5.6 family: Luna, Terra, Sol
  AI Unit Economics Evals Frontier Model Convergence / Commoditization Token Efficiency
• AI Root Cause Analysis Shifts from Model Reasoning to Context Engineering
  Evals Frontier Model Convergence / Commoditization Open-Weight Models
• The Almedra Token-Efficiency Benchmark, Updated For GLM 5.2
  AI Unit Economics Evals Token Efficiency
• Howard Yu's Post — American AI models went from 70% of usage to 30% in 12 months
  AI Unit Economics Frontier Model Convergence / Commoditization Open-Weight Models
• GitHub - musistudio/claude-code-router: Use Claude Code as the foundation for coding infrastructure, allowing you to decide how to interact with the model while enjoying updates from Anthropic.
  AI Unit Economics Frontier Model Convergence / Commoditization Open-Weight Models
• DeepSeek v4, and the end of the OpenAI/Microsoft AGI clause
  AI Unit Economics Evals Open-Weight Models
• Downstream #1: Where Do Our Tokens Go?
  AI Unit Economics Evals Open-Weight Models
• How Do AI Agents Spend Your Money? Analyzing and Predicting Token Consumption in Agentic Coding Tasks
  AI Unit Economics Evals Token Efficiency
• ChatGPT Just Became A Work Agent
  Evals Frontier Model Convergence / Commoditization
• AI Costs Are Surging and the Cheap Model Fix Might Not Last
  AI Unit Economics Open-Weight Models