Zhipu AI (Z.ai)Previous version · GLM-5.3

GLM-5.2

GLM-5.2 carries a "superseded" label, but its benchmark record is still the more independently verified of the two: eleven of its twelve tracked scores are independent runs, versus eight of ten for GLM-5.3, the model that replaced it. The one self-reported holdout is Humanity's Last Exam with tool use (54.7); on Terminal-Bench 2.1, Z.ai's 81.0 has given way to vals.ai's independent 67.79.

GLM-5.2 benchmarks and pricing, every number sourced: 12 tracked GLM-5.2 benchmark scores (11 independently run, 1 still resting on a vendor’s own claim), priced at $1.40 per million input tokens and $4.40 per million output.

GLM-5.2 architecture: Mixture-of-Experts with sparse attention; 1M-token context window.

Released
2026-06-16
License
open-weights
Context window
1M tokens
Knowledge cutoff
Not disclosed
Verified
sources checked 2026-06-30–2026-10-01
Parameters
Not disclosed
Architecture
Mixture-of-Experts with sparse attention

GLM-5.2’s verified record

GLM-5.2’s featured comparison is DeepSeek V4 Pro (0813): 2 trails, 1 tie, and 7 not callable. GLM-5.2 is priced at $1.40/$4.40 per 1M tokens (in/out) vs DeepSeek V4 Pro (0813)’s $1.32/$3.96. Full verdict →

Against the 270 head-to-head comparisons GLM-5.2 shares with other tracked models:

89 real gaps17 inside the band164 not called

A gap counts for GLM-5.2 only where both sides were run independently and the benchmark still separates models. Losses are listed alongside wins on purpose — GLM-5.2 trails on 79 of them.

HLE · no tools Reasoning

±2 is noise
Behind
Grok 4.7 −2.0 · DeepSeek V4.1 Flash −2.2 · Muse Spark Open −2.8 · Claude Haiku 5.5 −3.3 · Step 5 Preview −5.4 · Kimi K3 −5.8 · Gemini 3.1 Pro Preview −5.9 · Gemini 3.8 Flash −6.7 · MiniMax M3.1 Flash −6.9 · Muse Spark 1.3 −7.6 · MiMo-V2.6-Pro −8.3 · Muse Wonder −9.0 · Kimi K3.1 Pro −11.4 · Claude Sonnet 5.5 −13.9 · Gemini 4 Argon −16.0 · Claude Opus 5.5 −20.3 · 11 previous: Qwen3.8-Max −2.0 · Muse Spark 1.2 −4.4 · Gemini 3.7 Flash −6.8 · GPT-6 Sol −6.8 · Claude Opus 4.8 −7.6 · GPT-5.6 Sol −8.4 · GPT-6.1 Sol −11.8 · GPT-6 Astra −13.6 · Claude Opus 5 −13.8 · Claude Fable 5 −14.4 · Claude Fable 5.1 −18.0
Ahead
MiniMax M3 +2.1 · GPT-6 Luna +2.6 · Qwen3.8-Flash-Next +3.1 · Mistral Large 4 +6.1 · Beam 1.0 +6.4 · Nemotron 3.5 Super VL +10.4 · 1 previous: DeepSeek V4 Flash (0731) +2.5
Tie
6 models within ±2
Unverified
1 model — vendor-reported on one side

LiveBench Composite score across 7 domains

±2.7 is noise
Behind
GLM-5.3 −2.9 · Qwen3.8-Flash-Next −3.0 · Gemini 3.1 Pro Preview −3.8 · DeepSeek V4 Pro (0813) −4.2 · Grok 4.7 −4.2 · Claude Sonnet 5.5 −4.6 · Kimi K3 −6.0 · DeepSeek V4.1 Flash −7.9 · Muse Spark 1.3 −8.4 · Claude Opus 5.5 −10.0 · 12 previous: Claude Sonnet 5 −2.8 · Claude Opus 4.8 −3.0 · Grok 4.6 −4.8 · Muse Spark 1.2 −4.8 · Qwen3.8-Max −5.3 · Gemini 3.7 Flash −5.6 · GPT-6 Sol −6.1 · Claude Opus 5 −6.9 · GPT-5.6 Sol −7.8 · GPT-6.1 Sol −8.4 · Claude Fable 5 −9.8 · Claude Fable 5.1 −10.2
Ahead
Tie
6 models within ±2.7

DeepSWE Long-horizon coding

±9.5 is noise
Behind
GLM-5.3-Flash −19.0 · Kimi K3 −25.0 · GLM-5.3 −25.0 · Gemini 3.8 Flash −30.0 · 11 previous: Claude Sonnet 5 −10.0 · Muse Spark 1.2 −11.0 · Qwen3.8-Max −13.0 · Claude Opus 4.8 −15.0 · Gemini 3.7 Flash −21.0 · Grok 4.6 −23.0 · GPT-5.6 Luna −23.0 · Claude Fable 5 −26.0 · GPT-5.6 Sol −29.0 · Claude Opus 5 −30.0 · GPT-6 Astra −30.0
Tie
1 model within ±9.5
Unverified
19 models — vendor-reported on one side
Setup-dependent
2 models — scored on a different harness or effort tier

Toolathlon-Verified Multi-tool chores

±9.7 is noise
Behind
DeepSeek V4 Pro (0813) −14.5 · Kimi K3 −16.6 · GLM-5.3-Flash −18.5 · 4 previous: DeepSeek V4 Flash (0731) −10.8 · Claude Sonnet 5 −11.7 · Muse Spark 1.2 −16.0 · Claude Opus 4.8 −16.3
Tie
1 model within ±9.7
Unverified
11 models — vendor-reported on one side

Agents' Last Exam Professional work

±3.2 is noise
Behind
GPT-6 Luna −4.6 · Kimi K3 −7.9 · Claude Opus 5.5 −17.8 · 2 previous: Claude Opus 4.8 −6.6 · GPT-6 Astra −13.8
Unverified
7 models — vendor-reported on one side
Setup-dependent
9 models — scored on a different harness or effort tier

Terminal-Bench 2.1 Terminal ops

±10.6 is noise
Behind
Kimi K3 −13.1 · 1 previous: GPT-6 Sol −15.4
Ahead
Tie
3 models within ±10.6
Unverified
3 models — vendor-reported on one side
Setup-dependent
25 models — scored on a different harness or effort tier

ARC-AGI-2 · untiered Compositional visual reasoning

±9.2 is noise
Behind

No verdict for GLM-5.2 anywhere on GPQA Diamond, LiveCodeBench, SWE-bench Verified ( saturated); HLE · with tools ( nothing independently confirmed on both sides); HMMT ( contaminated).

GLM-5.2 API pricing

$1.40 in / $4.40 out per 1M tokens — official pricing source

What GLM-5.2 costs per job

GLM-5.2 cost for three reference workloads, computed from its list rates
WorkloadTokens in / outCost
One long chat turn100K / 10K$0.184
A codebase review1,000K / 100K$1.84
A day of agent work10,000K / 1,000K$18.40

Computed from GLM-5.2’s list rates above — cache discounts and batch tiers are not applied.

GLM-5.2 is one of 3 Zhipu AI (Z.ai) models tracked on this site, at these official list prices.

Zhipu AI (Z.ai) model pricing, official list rates
ModelIn / 1MOut / 1M
GLM-5.3-Flash$0.15$0.50
GLM-5.3$1.40$4.40
GLM-5.2(previous version)$1.40$4.40

GLM-5.2 benchmark scores

GLM-5.2 benchmark scores, provenance, and source links
BenchmarkScore
HLE(no tools)[1]
Reasoning · ±2 is noise
HLE(with tools)[2]
Reasoning · ±2 is noise
GPQA Diamondsaturated[3]
Expert science Q&A — not ranked at any gap size
Terminal-Bench 2.1[4]
Terminal ops · ±10.6 is noise
DeepSWE[5]
Long-horizon coding · ±9.5 is noise
Toolathlon-Verified
Multi-tool chores · ±9.7 is noise
Agents' Last Exam[6]
Professional work · ±3.2 is noise
SWE-bench Verifiedsaturated[7]
Bug fixing — not ranked at any gap size
LiveCodeBenchsaturated[8]
Contest coding — not ranked at any gap size
ARC-AGI-2(untiered)[9]
Compositional visual reasoning · ±9.2 is noise
LiveBench[10]
Composite score across 7 domains · ±2.7 is noise
HMMTcontaminated[11]
Competition mathematics — not ranked at any gap size

Who ran these numbers: 11 of 12 independent — vals.ai (4), artificialanalysis.ai (1), deepswe.datacurve.ai (1), toolathlon.xyz (1), snorkel.ai (1), arcprize.org (1), livebench.ai (1), matharena.ai (1); vendor self-reported (1).

  1. HLE: AA's own run, 'GLM-5.2 (max)' (text-only, no tools). Replaces Z.ai's self-reported 40.5. Z.ai's card reports 40.5 on the text-only subset — not stored as a row; this cell holds AA's independent run.
  2. HLE: Corroborated by https://docs.z.ai/guides/llm/glm-5.2.
  3. GPQA Diamond: vals.ai run, rank 42/133 (updated 2026-08-15). Replaces Z.ai's self-reported 91.2 — a 5.6-point gap between vendor claim and independent run. Z.ai's card reports 91.2 — not stored as a row; this cell holds vals.ai's 85.61.
  4. Terminal-Bench 2.1: vals.ai's own Terminus 2 run, board row 'GLM 5.2' (67.79 ± 0.99, reasoning effort max, pass@1), from its archived Terminal-Bench 2.1 table. Replaces Z.ai's self-reported 81.0 (Terminus-2 harness, Hugging Face README), 13.2 points higher. Z.ai's card reports 81.0 (Terminus-2) and 82.7 (best reported harness, Claude Code) — neither stored as a row; this cell holds vals.ai's 67.79.
  5. DeepSWE: Independent score (44% ± 2%, $3.92/task).
  6. Agents' Last Exam: Overall pass rate 20.4 (Claude Code, Max; score 41.1; $1,086). Corrects an earlier version of this page, which showed 41.1 — that was the partial-credit score, not the pass rate.
  7. SWE-bench Verified: vals.ai run, rank 14/83, bash-only harness, $0.71/test (updated 2026-08-14).
  8. LiveCodeBench: vals.ai run, rank 88/138 (updated 2026-08-15).
  9. ARC-AGI-2: GLM-5.2's official ARC-AGI-2 leaderboard row, dated 2026-06-13 on arcprize.org. No Low/Medium/High/Max split exists for this row — a single flat score.
  10. LiveBench: Board row "GLM-5.2" on the 2026-06-25 LiveBench release.
  11. HMMT: MathArena's leaderboard row is "GLM 5.2", 92.42% across all 33 HMMT Feb 2026 problems (live-verified 2026-08-25 at matharena.ai/?comp=hmmt--hmmt_feb_2026). Carries MathArena's own contamination-warning flag — "Model was released after competition release" — a disclosed risk that applies to every one of this site's 4 tracked HMMT Feb 2026 rows, GLM 5.2 included, since all 4 tracked models were released after HMMT Feb 2026 took place. Z.ai's card reports 92.5 (GPT-5.5-medium judge) — matches MathArena's independent 92.42 stored here.

Notes on the record

MIT-licensed weights on Hugging Face and ModelScope. Cached-input price is $0.26/1M, far below the $1.40 shown.

A previous version since GLM-5.3's API went live on 2026-08-19 at the identical price — which also satisfied the condition this entry pre-committed to ('keep current until GLM-5.3 is actually available via API'). The Flash-tier GLM-5.3-Flash shipped MIT-licensed weights on 2026-08-26, and GLM-5.3 itself followed on 2026-08-25 under a custom non-MIT license — MIT remains what distinguishes GLM-5.2's weights from the rest of the current GLM-5 line. It is also the GLM model with the most independent benchmark rows on this site: 11 of its 12 tracked scores are independent runs, versus 8 of 10 for GLM-5.3 (recounted 2026-10-08, when vals.ai's Terminal-Bench 4.0 row joined its set; earlier recounts noted above; earlier, on 2026-08-27, both models picked up independent rows through late August — GLM-5.3's share grew sharply on 2026-08-20 as Artificial Analysis and vals.ai ran three more of its benchmarks, and its DeepSWE row flipped from vendor self-report to that benchmark's official board on 2026-08-26 — but GLM-5.2 remains ahead on the raw independent-row count). GLM-5.3's weights appeared on Hugging Face on 2026-08-25, roughly the two weeks after the August 14 launch that Z.ai promised, with the safety review tied to its exploit-finding capability evidently passed (see the GLM-5.3 page for the release specifics).

No parameter count from Z.ai. The GLM-5.2 model card publishes an architecture but no total and no activated figure, so this page leaves both blank rather than borrowing one. HuggingFace's file widget computes 753B from the published weights, and Artificial Analysis lists 40B active; neither is a Z.ai statement, and the two answer different questions than a vendor spec would. Checked 2026-08-28.

Compare with

FAQ

Has GLM-5.2 been independently benchmarked?

Mostly, yes. Eleven of the twelve benchmark scores tracked for GLM-5.2 on this site come from independent runs — Artificial Analysis (HLE, no tools), vals.ai (GPQA Diamond, SWE-bench Verified, LiveCodeBench, Terminal-Bench 2.1), deepswe.datacurve.ai (DeepSWE), Toolathlon, Snorkel AI (Agents' Last Exam), ARC Prize (ARC-AGI-2), LiveBench's own board, and MathArena (HMMT Feb 2026). The one exception is Z.ai's own Humanity's Last Exam with tool use (54.7). Z.ai's Terminal-Bench 2.1 figure, 81.0, was replaced on 2026-10-01 by vals.ai's run, 67.79.

Is GLM-5.2 open source?

Yes. Z.ai published GLM-5.2's weights on Hugging Face and ModelScope under an MIT license. The contrast with its successor is real but no longer about downloadability: Z.ai shipped GLM-5.3's weights on 2026-08-25, as promised — under a custom "other" license rather than MIT. GLM-5.2's record and its MIT license stay live here.

Why is GLM-5.2 marked as a previous version if it's still usable?

Because this entry's own pre-set trigger condition fired, not because the model became unusable. GLM-5.2 was set to stay 'current' only until GLM-5.3 shipped with working API access — that happened on 2026-08-19, at an unchanged $1.40/$4.40 price, so the status flipped that day.

How much does the GLM-5.2 API cost?

$1.40 per million input tokens and $4.40 per million output tokens, per Z.ai's pricing page. Cached input is much cheaper at $0.26 per million tokens — roughly a fifth of the fresh-input rate — which matters for coding workflows that keep resending the same long context.

Is GLM-5.2 good for coding?

On the independently run benchmarks, its record holds up: 82.8 on SWE-bench Verified and 69.5 on LiveCodeBench, both vals.ai, plus 44% on DeepSWE's long-horizon coding tasks. Terminal-Bench 2.1 is the weak spot: Z.ai reported 81.0, but vals.ai's independent Terminus 2 run measured 67.79, 13.2 points lower.

What changed between GLM-5.2 and GLM-5.3?

Z.ai says GLM-5.3 uses the same base model as GLM-5.2, with the gains coming from scaled post-training rather than a new architecture. The API price didn't move — both list at $1.40/$4.40 per million tokens. Z.ai's own launch chart puts GLM-5.3's Terminal-Bench 2.1 score at 88.2 versus GLM-5.2's 81.0, but that comparison is vendor-to-vendor on both ends. Independent runs now exist for both, from different harnesses: vals.ai measured GLM-5.2 at 67.79, and Artificial Analysis measured GLM-5.3 at 83.9, 4.3 points under Z.ai's figure. On vals.ai's own board, which ran both, GLM-5.3 reads 71.54, 3.75 points above GLM-5.2 and inside the 10.6 band.

Further reading

Benchmark guides