Zhipu AI model product
GLM 5.2
2 is an open-source large language model from Zhipu AI with a 1M-token input context and 128K maximum output.
Updated Sep 8, 2026. Default version: GLM-5.2
LLMBoard Score77.0GLM-5.2
Coverage40%19 benchmark families
Context window1MTokens
Official input price$1.4Zhipu AI API
GLM 5.2 Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
GLM 5.2 Benchmark Results
Benchmark scores for GLM-5.2.
Sort by
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkAIME 2026 | Score99.20% | Rank01 | Participants22 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkCritPT | Score16.70% | Rank02 | Participants6 | Percentile80.00% | EvidenceC | Evaluated |
| BenchmarkIMO-AnswerBench | Score91.00% | Rank02 | Participants20 | Percentile94.74% | EvidenceC | Evaluated |
| BenchmarkProgram Bench | Score63.70% | Rank02 | Participants7 | Percentile83.33% | EvidenceC | Evaluated |
| BenchmarkFrontierSWE | Score74.00% | Rank05 | Participants16 | Percentile73.33% | EvidenceB | Evaluated |
| BenchmarkPostTrainBench | Score34.30% | Rank05 | Participants7 | Percentile33.33% | EvidenceC | Evaluated |
| BenchmarkSWE-Marathon | Score13.00% | Rank05 | Participants5 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkHMMT Feb 26 | Score92.50% | Rank06 | Participants12 | Percentile54.55% | EvidenceC | Evaluated |
| BenchmarkNL2Repo | Score48.90% | Rank08 | Participants22 | Percentile66.67% | EvidenceC | Evaluated |
| BenchmarkLM Arena Agent Praise Complaint | Score11.73% | Rank08 | Participants49 | Percentile85.42% | EvidenceA | Evaluated |
| BenchmarkHMMT 2025 | Score94.40% | Rank09 | Participants33 | Percentile75.00% | EvidenceC | Evaluated |
| BenchmarkLM Arena Agent Tool Hallucination | Score0.78% | Rank09 | Participants49 | Percentile83.33% | EvidenceA | Evaluated |
| BenchmarkDeepSWE | Score46.20% | Rank10 | Participants13 | Percentile25.00% | EvidenceC | Evaluated |
| BenchmarkMCP Atlas | Score76.80% | Rank11 | Participants34 | Percentile69.70% | EvidenceC | Evaluated |
| BenchmarkLM Arena Agent Leaderboard | Score6.03% | Rank11 | Participants49 | Percentile79.17% | EvidenceA | Evaluated |
| BenchmarkLM Arena Agent Task Outcome Explicit | Score8.63% | Rank12 | Participants49 | Percentile77.08% | EvidenceA | Evaluated |
| BenchmarkLM Arena Agent Steerability | Score4.48% | Rank13 | Participants49 | Percentile75.00% | EvidenceA | Evaluated |
| BenchmarkFrontierCode 1.1 | Score24.50% | Rank14 | Participants17 | Percentile18.75% | EvidenceB | Evaluated |
| BenchmarkSWE-Bench Pro | Score62.10% | Rank14 | Participants55 | Percentile75.93% | EvidenceC | Evaluated |
| BenchmarkHumanity's Last Exam | Score54.70% | Rank16 | Participants103 | Percentile85.29% | EvidenceC | Evaluated |
| BenchmarkLM Arena Webdev | Score1,586.57 rating | Rank16 | Participants97 | Percentile84.38% | EvidenceA | Evaluated |
| BenchmarkTerminal-Bench 2.1 | Score82.70% | Rank18 | Participants35 | Percentile50.00% | EvidenceC | Evaluated |
| BenchmarkGPQA | Score91.20% | Rank22 | Participants247 | Percentile91.46% | EvidenceC | Evaluated |
| BenchmarkLM Arena Agent Bash Recovery Steps | Score4.51% | Rank22 | Participants49 | Percentile56.25% | EvidenceA | Evaluated |
| BenchmarkLM Arena Text | Score1,466.70 rating | Rank23 | Participants210 | Percentile89.47% | EvidenceA | Evaluated |
| BenchmarkDeepSWE 1.1 | Score44.00% | Rank25 | Participants31 | Percentile20.00% | EvidenceB | Evaluated |
| BenchmarkToolathlon | Score48.20% | Rank27 | Participants41 | Percentile35.00% | EvidenceC | Evaluated |
| BenchmarkLM Arena Text Factuality | Score1,461.04 rating | Rank29 | Participants121 | Percentile76.67% | EvidenceA | Evaluated |
| BenchmarkLM Arena Text Style Control | Score1,471.72 rating | Rank30 | Participants210 | Percentile86.12% | EvidenceA | Evaluated |
GLM 5.2 Arena Results
Preference and agent-evaluation results for the default version.
Sort by
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenatext factuality | Categoryspanish | Rank05 | Rating / score1,482.40 | Votes747 | ObservationsN/A | Result date |
| Arenatext | Categoryspanish | Rank06 | Rating / score1,481.08 | Votes1,003 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryjapanese | Rank07 | Rating / score1,455.67 | Votes319 | ObservationsN/A | Result date |
| Arenaagent praise complaint | Categoryoverall | Rank08 | Rating / score0.12 | VotesN/A | Observations15.4K | Result date |
| Arenaagent tool hallucination | Categoryoverall | Rank09 | Rating / score0.01 | VotesN/A | Observations3.7M | Result date |
| Arenatext factuality | Categorykorean | Rank10 | Rating / score1,432.28 | Votes365 | ObservationsN/A | Result date |
| Arenaagent | Categoryoverall | Rank11 | Rating / score0.06 | VotesN/A | Observations4M | Result date |
| Arenaagent task outcome explicit | Categoryoverall | Rank12 | Rating / score0.09 | VotesN/A | Observations42.5K | Result date |
| Arenatext style control | Categoryspanish | Rank12 | Rating / score1,479.80 | Votes1,003 | ObservationsN/A | Result date |
| Arenaagent steerability | Categoryoverall | Rank13 | Rating / score0.04 | VotesN/A | Observations52.5K | Result date |
| Arenatext | Categorycreative writing | Rank14 | Rating / score1,462.38 | Votes6,493 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryindustry medicine and healthcare | Rank15 | Rating / score1,495.35 | Votes2,041 | ObservationsN/A | Result date |
| Arenawebdev | Categorywebdev-react | Rank15 | Rating / score1,598.59 | Votes6,950 | ObservationsN/A | Result date |
| Arenatext | Categorykorean | Rank16 | Rating / score1,435.69 | Votes682 | ObservationsN/A | Result date |
| Arenatext factuality | Categorycreative writing | Rank16 | Rating / score1,455.70 | Votes6,048 | ObservationsN/A | Result date |
| Arenawebdev | Categoryoverall | Rank16 | Rating / score1,586.57 | Votes9,836 | ObservationsN/A | Result date |
| Arenawebdev | Categorywebdev | Rank16 | Rating / score1,586.57 | Votes9,836 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry life and physical and social science | Rank19 | Rating / score1,481.10 | Votes5,547 | ObservationsN/A | Result date |
| Arenatext | Categoryjapanese | Rank19 | Rating / score1,446.19 | Votes505 | ObservationsN/A | Result date |
| Arenatext | Categorychinese | Rank20 | Rating / score1,517.30 | Votes2,088 | ObservationsN/A | Result date |
| Arenatext | Categorygerman | Rank20 | Rating / score1,465.50 | Votes618 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry entertainment and sports and media | Rank20 | Rating / score1,444.99 | Votes8,275 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry medicine and healthcare | Rank20 | Rating / score1,471.24 | Votes2,461 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry writing and literature and language | Rank20 | Rating / score1,459.03 | Votes8,888 | ObservationsN/A | Result date |
| Arenatext factuality | Categorychinese | Rank20 | Rating / score1,512.30 | Votes1,755 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryindustry entertainment and sports and media | Rank20 | Rating / score1,446.68 | Votes8,175 | ObservationsN/A | Result date |
| Arenatext style control | Categoryjapanese | Rank20 | Rating / score1,452.99 | Votes505 | ObservationsN/A | Result date |
| Arenatext style control | Categorykorean | Rank20 | Rating / score1,436.69 | Votes682 | ObservationsN/A | Result date |
| Arenatext | Categoryenglish | Rank21 | Rating / score1,471.80 | Votes14,561 | ObservationsN/A | Result date |
| Arenaagent bash recovery steps | Categoryoverall | Rank22 | Rating / score0.05 | VotesN/A | Observations116.8K | Result date |
GLM 5.2 Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- $1.4 input, $4.4 output per 1M
- Official provider
- Zhipu AI
- Lowest third-party
- From $0.30 input, $1.05 output per 1M via CrofAI
- Tracked offerings
- 89
Sort by
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderUmans AI Coding Plan | Provider model IDumans-glm-5.2 | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context405.5K | Updated |
| ProviderNvidia | Provider model IDz-ai/glm-5.2 | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context1M | Updated |
| ProviderAlibaba Token Plan | Provider model IDglm-5.2 | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context1M | Updated |
| ProviderKenari | Provider model IDglm-5-2 | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context1M | Updated |
| ProviderSenseNova (China) | Provider model IDglm-5.2 | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context1M | Updated |
| ProviderUnoRouter | Provider model IDglm-5.2:free | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context1M | Updated |
| ProviderZhipu AI Coding Plan | Provider model IDglm-5.2 | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context1M | Updated |
| ProviderZ.AI Coding Plan | Provider model IDglm-5.2 | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context1M | Updated |
| ProviderSCNet Token Plan | Provider model IDGLM-5.2 | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context1M | Updated |
| ProviderAlibaba Token Plan (China) | Provider model IDglm-5.2 | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context1M | Updated |
| ProviderNaN | Provider model IDglm5.2 | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context500K | Updated |
| ProviderCrofAI | Provider model IDglm-5.2 | Regionglobal | Input / 1M$0.30 | Output / 1M$1.05 | Context1M | Updated |
| ProviderNanoGPT | Provider model IDz-ai/glm-5.2 | Regionglobal | Input / 1M$0.42 | Output / 1M$1.32 | Context1M | Updated |
| ProviderSCX.ai | Provider model IDGLM-5.2 | Regionglobal | Input / 1M$0.55 | Output / 1M$1.78 | Context1M | Updated |
| ProviderDevPass (LLM Gateway) | Provider model IDglm-5.2 | Regionglobal | Input / 1M$0.55 | Output / 1M$1.78 | Context1M | Updated |
| ProviderAmbient | Provider model IDz-ai/glm-5.2 | Regionglobal | Input / 1M$0.60 | Output / 1M$2 | Context202.8K | Updated |
| ProviderDigitalOcean | Provider model IDglm-5.2 | Regionglobal | Input / 1M$0.70 | Output / 1M$2.2 | Context262.1K | Updated |
| ProviderInceptron | Provider model IDzai-org/GLM-5.2 | Regionglobal | Input / 1M$0.71 | Output / 1M$2.35 | Context1M | Updated |
| ProviderDeep Infra | Provider model IDzai-org/GLM-5.2 | Regionglobal | Input / 1M$0.75 | Output / 1M$2.4 | Context1M | Updated |
| ProviderWeights & Biases | Provider model IDzai-org/GLM-5.2 | Regionglobal | Input / 1M$0.76 | Output / 1M$2.42 | Context1M | Updated |
| Providerrouting.run | Provider model IDglm-5.2 | Regionglobal | Input / 1M$0.80 | Output / 1M$2.4 | Context200K | Updated |
| ProviderRequesty | Provider model IDglm-5.2 | Regionglobal | Input / 1M$0.80 | Output / 1M$2.55 | Context1M | Updated |
| ProviderVercel AI Gateway | Provider model IDzai/glm-5.2 | Regionglobal | Input / 1M$0.80 | Output / 1M$2.55 | Context1M | Updated |
| ProviderVultr | Provider model IDzai-org/GLM-5.2-FP8 | Regionglobal | Input / 1M$0.85 | Output / 1M$3.1 | Context393.2K | Updated |
| ProviderLilac | Provider model IDzai-org/glm-5.2 | Regionglobal | Input / 1M$0.90 | Output / 1M$3 | Context524.3K | Updated |
| ProviderOpenRouter | Provider model IDz-ai/glm-5.2 | Regionglobal | Input / 1M$0.966 | Output / 1M$3.04 | Context1M | Updated |
| ProviderGMI Cloud | Provider model IDzai-org/GLM-5.2-FP8 | Regionglobal | Input / 1M$0.979 | Output / 1M$3.08 | Context1M | Updated |
| ProviderOfox | Provider model IDz-ai/glm-5.2 | Regionglobal | Input / 1M$0.98 | Output / 1M$3.08 | Context1M | Updated |
| Providerai& | Provider model IDzai-org/glm-5.2 | Regionglobal | Input / 1M$1 | Output / 1M$4 | Context1M | Updated |
| ProviderMerge Gateway | Provider model IDzai/glm-5.2 | Regionglobal | Input / 1M$1.05 | Output / 1M$3.3 | Context1M | Updated |
GLM 5.2 Runtime Performance
Provider-specific output speed and catalog latency for GLM-5.2. Runtime does not affect the capability score.
Sort by
Provider | Output Speed | Catalog Latency | Max Input | Max Output | Updated |
|---|
| ProviderZAI | Output Speed3.96 tok/s | Catalog Latency14.29 s | Max Input1M | Max Output128K | Updated |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
GLM 5.2 Specifications
Technical details for the model's default version.
- Version
- GLM-5.2
- Released
- Jun 16, 2026
- Knowledge cutoff
- Unknown
- Parameters
- 753B
- Context window
- 1M
- Max output
- 131.1K
- Inputs
- text
- Outputs
- text
- Open weights
- Yes
- License
- MIT
GLM 5.2 Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
Sort by
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionGLM-5.2 | Released | LLMBoard77.02 | Parameters753B | Context1M | Max output131.1K | Open weightsYes | LicenseMIT |
Models similar to GLM 5.2
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
ZA
GLM 5.3 Flash
Zhipu AI
ZA
GLM 5.1
Zhipu AI
ZA
GLM 5
Zhipu AI
ZA
GLM 5.3
Zhipu AI
ZA
GLM 4.7
Zhipu AI
ZA
GLM 5V Turbo
Zhipu AI
What is GLM 5.2?
Key information about GLM 5.2 and its available data.
2 is Zhipu AI's foundation model for long-horizon tasks, including coding-agent workloads. It is released under the MIT license and supports thinking mode, function calling, structured output, context caching, and MCP integration.
Data as of 2026-09-08.
FAQ
Common questions about GLM 5.2.
When was GLM 5.2 released?
GLM 5.2's default version was released on Jun 16, 2026.
How much does GLM 5.2 cost?
GLM 5.2's official API price is $1.4 per million input tokens and $4.4 per million output tokens via Zhipu AI. The lowest tracked third-party offer starts at $0.30 input and $1.05 output via CrofAI.
Who created GLM 5.2?
GLM 5.2 was created by Zhipu AI.
What is the context window for GLM 5.2?
The default version has a 1M token context window.
Is GLM 5.2 open weight?
Yes. The default version is marked as open weight under MIT.
How many API providers offer GLM 5.2?
89 provider offerings are linked to the default version.