BenchLeader

GLM-5

Reasoning effort
Quality#39 of 337
60.2Index
Top quarter
+10.3 on the median
Price#161 of 253
$1.55per 1M tokens
Dearer than most
1.8× the median
Speed#145 of 215
62tokens/s
Below the median
0.8× the median
Value#156 of 253
38.8points per $
Below the median
0.7× the median

Better value: GPT-5.6 Luna scores 58.2 for $0.450/M, 71% cheaper.

3 cheaper ranked models score higher: Gemini 3.7 Flash (62.1, $1.50), Gemini 3.8 Flash (61.6, $1.50), Gemini 3.6 Flash (61.0, $1.50).

As of 15 Nov 2026, GLM-5 by Zhipu AI scores 60.2 on the BenchLeader Index (top 100 of 607 ranked configurations), with its best results in instruction following (71) and weakest in knowledge (57). API pricing: $1.00 per 1M input tokens and $3.20 per 1M output tokens ($1.55 blended at 3:1), list price on 15 Nov 2026. It streams about 62 tokens/s (Artificial Analysis). Released 11 Feb 2026.

Key facts (as of 15 Nov 2026)

  • GLM-5 API pricing: $1.00 per 1M input tokens and $3.20 per 1M output tokens ($1.55 blended at 3:1) (list price, 15 Nov 2026).
  • GLM-5 streams about 62 tokens/s (Artificial Analysis, 15 Nov 2026).
  • GLM-5 was released on 11 Feb 2026 (Zhipu AI).

In detail. GLM-5 is a Zhipu AI open-weights reasoning model, released 11 Feb 2026. Its best configuration (thinking reasoning effort) ranks #84 of 607 on the BenchLeader Index at 60.2 ±3.6, in the upper half. The ± is the point: 201 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in instruction following (71) and lowest in knowledge (57). At $1.55 per million tokens blended it is pricier than most ranked models. Output speed of 62 tokens per second puts it slower than most, with a first answer in 34.3 s. It has been measured at 4 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 13 Oct 2026.

Blended price
$1.55/M
$1.00 in · $3.20 out
via Artificial Analysis
Output speed
62 tok/s
measured by Artificial Analysis
First answer
34 s
first token 1.82 s
Context
205k
via models.dev
Full answer
81 s
last measured 13 Nov 2026
Cost per task
–
Tokens per task
–
Released
11 Feb 2026
How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.
  1. Overall index60
  2. Reasoning68
  3. Coding59
  4. Agents & tools58
  5. Maths63
  6. Knowledge57
  7. Instruction following71
  8. Long context62

Embed this rank

More badge styles
[![GLM-5 on BenchLeader: #39 overall](https://www.benchleader.com/badge/glm-5.svg?style=compact)](https://www.benchleader.com/models/glm-5)

Paste it at the top of README.md, beside your other badges. The badge redraws with every data update and links back to this page.

Get an email when Zhipu AI ships a new model

And once when it gets ranked, plus the weekly review on Mondays.

Choose providers and emails

Versions

Zhipu AI has shipped 8 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.

ModelReleasedIndexRank
GLM 5.3max18 Aug 202658.8#118
GLM 5.2max16 Jun 202659.0#112
GLM 5.17 Apr 202655.4#206
GLM-5thinkingthis page11 Feb 202660.2#84
GLM 4.722 Dec 202551.7#303
GLM-4.6thinking30 Sep 202551.4#316
GLM-4.528 Jul 202551.6#308
GLM-4-0520–––

Reasoning-effort configurations

The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.

EffortIndexRankSpeedFirst answerChat reply costCategories
no reasoning––53 tok/s2.34 s$0.0014Agents & tools 68 · Instruction following 56 · Long context 46
high––43 tok/s1.82 s$0.0014Coding 64
thinkingbest60.2#8462 tok/s34 s$0.0014Agents & tools 58 · Coding 59 · Instruction following 71 · Knowledge 57 · Long context 62 · Maths 63 · Reasoning 68
not stated55.0#21743 tok/s1.82 s$0.0014Agents & tools 59 · Coding 51 · Human preference 64 · Maths 55 · Reasoning 50

How its index has moved

8 Sep 2026 to 15 Nov 2026
515864
  • thinking
  • not stated

The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.

Benchmark results

See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Metadata

Benchmarkno reasoningthinkingSourceTrend
AA Openness Indexnot in index50#6150#61Artificial Analysis

Reasoning

Benchmarkno reasoningthinkingnot statedSourceTrend
GPQA Diamondeffort not stated––87.8%#81Epoch AI Benchmarking Hub
SimpleBencheffort not stated––53.2%#57SimpleBench
Arena Hard Promptseffort not stated––1478#71Arena (LMArena)
GPQA Diamond (AA)not in index66.6%#33682.0%#178–Artificial Analysis
Humanity's Last Exam (AA)not in index7.7%#37829.3%#173–Artificial Analysis
GPQA Diamond (Vals)not in index–83.3%#57–Vals AI
Kagi LLM Benchmark–75.0%#1251.7%#81Kagi LLM Benchmark
ARC-AGI-1effort not stated––44.7%#197ARC Prize
ARC-AGI-2effort not stated––4.9%#187ARC Prize
CritPtnot in index0.0%#3372.0%#220–Artificial Analysis
CL-benchnot in indexeffort not stated––18.7%#11Epoch AI Benchmarking Hub
ForecastBenchnot in indexeffort not stated––63.4%#42ForecastBench

Coding

Benchmarkhighthinkingnot statedSourceTrend
SWE-bench Verified (Epoch)effort not stated––72.1%#22Epoch AI Benchmarking Hub
WeirdMLeffort not stated––48.2%#73WeirdML
Arena Codingeffort not stated––1498#86Arena (LMArena)
Arena WebDeveffort not stated––1435#90Arena (LMArena)
LiveCodeBench–81.9%#58–Vals AI
SWE-bench (Vals)not in index–71.4%#56–Vals AI
SWE-bench Verified (bash only)72.8%#7––SWE-bench
SWE-bench Verified (any scaffold)not in index72.8%#10––SWE-bench
Vibe Code Bench v1.1not in index–23.4%#88–Vals AI
SWE Atlas: Codebase QnAnot in indexeffort not stated––20.5%#18Scale AI SEAL
SWE Atlas: Refactoringnot in indexeffort not stated––24.2%#14Scale AI SEAL
SWE Atlas: Test Writingnot in indexeffort not stated––28.7%#19Scale AI SEAL
ALE-Benchnot in indexeffort not stated––765.6#91Epoch AI Benchmarking Hub
Design Arena Websitenot in indexeffort not stated––1239#68Design Arena

Agents & tools

Benchmarkno reasoningthinkingnot statedSourceTrend
Terminal-Bencheffort not stated––52.4%#36Terminal-Bench
Terminal-Bench Hard39.4%#5743.2%#41–Artificial Analysis
τ²-Bench Telecom (AA)not in index97.4%#1298.3%#8–Artificial Analysis
τ²-bench–81.0%#6–τ²-bench
τ³-bench Bankingnot in index–9.8%#26–τ²-bench
APEX-Agents (AA)not in index–14.4%#25–Artificial Analysis
Terminal-Bench 2.0 (Vals)not in index–49.4%#21–Vals AI
Vending-Bench 2not in indexeffort not stated––4432.1#41Andon Labs
Design Arena Agenticnot in indexeffort not stated––1114#55Design Arena

Maths

Benchmarkthinkingnot statedSourceTrend
OTIS Mock AIMEeffort not stated–80.0%#125Epoch AI Benchmarking Hub
AIME (Vals)91.7%#23–Vals AI
AIME 2026effort not statedpossible contamination–96.7%#6MathArena
HMMT February 2026effort not stated–86.4%#20MathArena
USAMO 2026not in indexeffort not stated–35.1%#9MathArena
MathArena Apexeffort not statedpossible contamination–10.9%#21MathArena
Arena Mathseffort not stated–1443#89Arena (LMArena)

Knowledge

Benchmarkno reasoningthinkingSourceTrend
MMLU-Pro–86.0%#50Vals AI
LegalBench–84.1%#42Vals AI
CorpFin–62.9%#48Vals AI
TaxEval–70.0%#94Vals AI
MedQA–94.3%#18Vals AI
CaseLaw v2not in index–52.5%#45Vals AI
Global-MMLU-Lite (AA)not in index81.9%#58–Artificial Analysis

Instruction following

Benchmarkno reasoningthinkingnot statedSourceTrend
IFBench55.2%#14872.3%#49–Artificial Analysis
Arena Instruction Followingnot in indexeffort not stated––1447#72Arena (LMArena)

Human preference

Benchmarknot statedSourceTrend
Arena Text1457#67Arena (LMArena)
Arena Creative Writingnot in index1447#48Arena (LMArena)
Arena Multi-turnnot in index1472#59Arena (LMArena)
Arena Longer Queriesnot in index1469#66Arena (LMArena)

Long context

Benchmarkno reasoningthinkingSourceTrend
AA-LCR v1.143.7%#36675.7%#171Artificial Analysis

Composite

Benchmarkno reasoningthinkingnot statedSourceTrend
Epoch Capabilities Indexnot in indexeffort not stated––146.5#84Epoch AI Benchmarking Hub
AA Intelligence Index v4.3.2not in index21.8#21827.9#146–Artificial Analysis
AA Progress Index v1not in index246.1#275280.9#176–Artificial Analysis

API pricing: where to run it

Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.

ProviderSpeedFirst tokenInput $/MOutput $/MContextQuantisation
Amazon Bedrock118 tok/s0.69 s$1.00$3.20203k–
Baidu Qianfan62 tok/s1.70 s$0.700$2.24203kfp8
Venice (fp8)57 tok/s1.74 s$1.00$3.20198kfp8
GMICloud47 tok/s2.15 s$0.600$1.92203kfp8
Z.ai43 tok/s1.73 s$1.00$3.20203kfp8
NovitaAI43 tok/s2.46 s$1.00$3.20203kfp8
StreamLake39 tok/s1.86 s$0.600$1.92198kfp8
SiliconFlow30 tok/s3.14 s$0.950$2.55205kfp8
Venice28 tok/s4.18 s$1.20$4.00200k–

What a task costs

Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.200 per 1M. Reasoning tokens are not modelled.

WorkloadTokens in / outCostWith cachingTime
Chat reply400 / 300$0.0014$0.001139.1 s
Summarize a 30-page report12,000 / 600$0.014$0.006743.9 s
Code edit6,000 / 1,500$0.011$0.007258.3 s
Agentic coding session60,000 / 4,000$0.073$0.0371.6 min
Structured extraction2,000 / 200$0.0026$0.001437.5 s

More from Zhipu AI

New this month

Compare GLM-5 with…

See also

Embed or cite

Show where GLM-5 ranks on a model card or README. The badge redraws itself with every daily update, carries the data date and links back to this page.

[![GLM-5 on BenchLeader: #39 overall](https://www.benchleader.com/badge/glm-5.svg)](https://www.benchleader.com/models/glm-5)

Cite this

The ranking changes daily, so a reference carries the data date. APA style:

BenchLeader. (2026, November 15). GLM-5 on the BenchLeader Index (method v1.4) [Data set]. Retrieved November 15, 2026, from https://www.benchleader.com/models/glm-5

BibTeX:

@misc{benchleader_glm_5_20261115,
  author       = {{BenchLeader}},
  title        = {{GLM-5 on the BenchLeader Index (method v1.4)}},
  year         = {2026},
  month        = nov,
  howpublished = {\url{https://www.benchleader.com/models/glm-5}},
  note         = {Data as of 2026-11-15}
}

Every badge style, the embeddable top-10 widget and how to cite the Index

Questions people ask

What are GLM-5's benchmark scores?
As of 15 Nov 2026, GLM-5 (thinking effort) scores 60.2 ±3.6 on the BenchLeader Index, top 100, strongest in instruction following (71) and weakest in knowledge (57). That rests on independent benchmark results only, and every score on this page links to its publisher.
How does GLM-5 rank?
GLM-5 (thinking effort) is #84 of 607 ranked configurations on 15 Nov 2026 (top 100). 201 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Fable 5.2 (max effort) at 72.6.
How much does GLM-5 cost per 1M tokens?
As of 15 Nov 2026, GLM-5 costs $1.00 per 1M input tokens and $3.20 per 1M output tokens, $1.55 blended (three input tokens per output token).
Is GLM-5 better than Gemini 3 Pro?
As of 15 Nov 2026, Gemini 3 Pro and GLM-5 (thinking) are neck and neck on quality (60.1 vs 60.2, inside the ±5.0 margin of error) and close on price, so pick on the category that matters to you. The head-to-head page has the benchmark-by-benchmark detail.
When was GLM-5 released?
Zhipu AI released GLM-5 on 11 Feb 2026. BenchLeader has results for it at 4 reasoning-effort settings.

Data as of 15 Nov 2026. Compare these configurations.

Cite as: BenchLeader, “GLM-5: benchmarks, pricing, speed and rank”, data as of 15 Nov 2026, https://www.benchleader.com/models/glm-5.