Zhipu AIGLM 5.3
GLM 5.3 is a Zhipu AI open-weights reasoning model, released 18 Aug 2026. Its best configuration (max reasoning effort) ranks #63 of 785 on the BenchLeader Index at 63.5 ±3.2, in the upper half. The ± is the point: 136 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 composite (80) and lowest in agents & tools (58). At $2.15 per million tokens blended it is pricier than most ranked models. Output speed of 85 tokens per second puts it slower than most, with a first answer in 26.5 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 22 Oct 2026.
- Blended price
- $2.15/M
- $1.40 in · $4.40 out
- Output speed
- 85 tok/s
- measured by Artificial Analysis
- First answer
- 27 s
- first token 0.83 s
- Context
- 1M
- Full answer
- 32 s
- median, reasoning included
- Cost per run
- $2.01
- one full Intelligence Index run
- Released
- 18 Aug 2026
- Overall index64
- Reasoning69
- Coding62
- Agents & tools58
- Maths60
- Knowledge59
- Human preference67
- Long context65
- Composite80
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.
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.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| low | 55.6 | #226 | 90 tok/s | 25 s | $0.0019 | Coding 43 · Composite 68 · Knowledge 58 · Long context 61 · Reasoning 65 |
| high | – | – | 103 tok/s | 0.83 s | $0.0019 | |
| maxbest | 63.5 | #63 | 85 tok/s | 27 s | $0.0019 | Agents & tools 58 · Coding 62 · Composite 80 · Human preference 67 · Knowledge 59 · Long context 65 · Maths 60 · Reasoning 69 |
| not stated | – | – | 103 tok/s | 0.83 s | $0.0019 | Agents & tools 62 · Composite 42 |
What more thinking costs
Turning the effort up buys index points and multiplies the bill. Here max costs 2.4× low for +7.9 on the index — 6.4 points per doubling of spend. Compare that with other models.
How its index has moved
8 Sept 2026 to 23 Oct 2026- max
- low
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
| Benchmark | max | Source | Trend |
|---|---|---|---|
| AA Openness Indexnot in index | 33.3#34 | Artificial Analysis |
Reasoning
| Benchmark | low | max | not stated | Source | Trend |
|---|---|---|---|---|---|
| GPQA Diamond | – | 90.9%#35 | – | Epoch AI Benchmarking Hub | |
| Arena Hard Prompts | – | 1505#23 | – | Arena (LMArena) | |
| LiveBench 2.0 Reasoningnot in indexeffort not stated | – | – | 85.8%#29 | LiveBench | |
| LiveBench 1.0 Reasoningnot in indexeffort not stated | – | – | 85.8%#35 | LiveBench | |
| GPQA Diamond (AA)not in index | – | 91.7%#45 | – | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 36.6%#111 | 42.3%#73 | – | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | – | 88.1%#34 | – | Vals AI | |
| CritPt | 14.6%#97 | 19.1%#67 | – | Artificial Analysis | |
| MysteryMechanismnot in index | – | 23.0%#18 | – | Vals AI | |
| Chess Puzzlesnot in index | – | 37.0%#44 | – | Epoch AI Benchmarking Hub | |
| Mystery Game Puzzlesnot in index | – | 33.0%#28 | – | Epoch AI Benchmarking Hub | |
| LMCAnot in indexeffort not stated | – | – | 55.5%#45 | Epoch AI Benchmarking Hub | |
| DTBenchnot in indexeffort not stated | – | – | 87.7%#99 | Epoch AI Benchmarking Hub | |
| BTF-3not in indexeffort not stated | – | – | 14.9%#1 | Epoch AI Benchmarking Hub |
Coding
Agents & tools
Maths
| Benchmark | max | not stated | Source | Trend |
|---|---|---|---|---|
| FrontierMath Tiers 1–3 | 68.8%#30 | – | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4 | 29.3%#40 | – | Epoch AI Benchmarking Hub | |
| OTIS Mock AIME | 91.1%#70 | – | Epoch AI Benchmarking Hub | |
| ProofBench | 49.0%#33 | – | Vals AI | |
| LiveBench Mathematicsnot in indexeffort not stated | – | 87.9%#44 | LiveBench | |
| Arena Maths | 1500#18 | – | Arena (LMArena) |
Knowledge
| Benchmark | low | max | not stated | Source | Trend |
|---|---|---|---|---|---|
| SimpleQA Verified | – | 41.0%#46 | – | Epoch AI Benchmarking Hub | |
| LiveBench 2.0 Data Analysisnot in indexeffort not stated | – | – | 70.2%#36 | LiveBench | |
| LiveBench Researchnot in indexeffort not stated | – | – | 73.2%#26 | LiveBench | |
| LiveBench 1.0 Data Analysisnot in indexeffort not stated | – | – | 70.2%#51 | LiveBench | |
| AA-Omniscience | -8.4#180 | 14.3#85 | – | Artificial Analysis | |
| MMLU-Pro | – | 86.8%#37 | – | Vals AI | |
| LegalBench | – | 84.8%#29 | – | Vals AI | |
| TaxEval | – | 72.4%#64 | – | Vals AI | |
| Excel Modeling Benchmarknot in index | – | 56.3%#46 | – | Vals AI | |
| Excel Modeling Benchmark v1.1not in index | – | 69.1%#16 | – | Vals AI | |
| MedCodenot in index | – | 42.9%#48 | – | Vals AI | |
| MedScribenot in index | – | 88.8%#11 | – | Vals AI | |
| Public Benefits Bench v1.1not in index | – | 68.5%#9 | – | Vals AI | |
| AA-Omniscience: accuracynot in index | 33.9%#161 | 33.9%#162 | – | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 36.0%#184 | 70.5%#40 | – | Artificial Analysis |
Instruction following
| Benchmark | max | not stated | Source | Trend |
|---|---|---|---|---|
| LiveBench Languagenot in indexeffort not stated | – | 79.9%#31 | LiveBench | |
| LiveBench Instruction Followingnot in indexeffort not stated | – | 69.3%#32 | LiveBench | |
| Arena Instruction Followingnot in index | 1478#22 | – | Arena (LMArena) |
Human preference
| Benchmark | max | Source | Trend |
|---|---|---|---|
| Arena Text | 1481#26 | Arena (LMArena) | |
| Arena Creative Writingnot in index | 1456#31 | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1478#43 | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1486#35 | Arena (LMArena) |
Multimodal
| Benchmark | max | not stated | Source | Trend |
|---|---|---|---|---|
| LiveBench Designnot in indexeffort not stated | – | 77.3%#22 | LiveBench | |
| AA-3Dnot in index | 787#19 | – | Artificial Analysis |
Long context
| Benchmark | low | max | Source | Trend |
|---|---|---|---|---|
| AA-LCR | 72.7%#187 | 79.7%#88 | Artificial Analysis | |
| MLCRnot in index | 11.1%#104 | 17.2%#65 | Artificial Analysis |
Composite
| Benchmark | low | max | not stated | Source | Trend |
|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in indexeffort not stated | – | – | 156.0#31 | Epoch AI Benchmarking Hub | |
| LiveBench 2.0effort not stated | – | – | 60.7%#39 | LiveBench | |
| LiveBench 1.0not in indexeffort not stated | – | – | 76.1%#32 | LiveBench | |
| AA Intelligence Index v4.3.2 | 34.3#96 | 44.8#40 | – | Artificial Analysis | |
| Vals Indexnot in index | – | 59.7#9 | – | Vals AI | |
| Vals Finance Indexnot in index | – | 60.5%#8 | – | Vals AI | |
| Vals RSI Indexnot in index | – | 22.9%#13 | – | Vals AI | |
| Vals Router Indexnot in indexeffort not stated | – | – | 59.7%#9 | Vals AI | |
| AA Cyber Indexnot in index | – | 36.4#12 | – | Artificial Analysis |
Safety & honesty
| Benchmark | max | Source | Trend |
|---|---|---|---|
| FORTRESSnot in index | 28.2%#26 | Scale AI SEAL |
Where it wins
Benchmarks where this configuration ranks in the top five of every configuration measured.
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.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| Baseten (fp8) | 144 tok/s | 0.86 s | $2.10 | $6.60 | 1.0M | fp8 |
| Mistral | 142 tok/s | 1.21 s | $1.40 | $4.40 | 1.0M | nvfp4 |
| Baseten (fp8) | 135 tok/s | 1.57 s | $2.10 | $6.60 | 1.0M | fp8 |
| Mistral | 126 tok/s | 0.80 s | $1.54 | $4.84 | 1.0M | nvfp4 |
| inference.net | 117 tok/s | 0.56 s | $0.080 | $4.00 | 1.0M | fp4 |
| Prime Intellect | 114 tok/s | 1.19 s | $0.980 | $3.08 | 1.0M | – |
| inference.net | 113 tok/s | 0.72 s | $0.080 | $4.38 | 1.0M | fp4 |
| Crusoe | 113 tok/s | 0.72 s | $1.40 | $4.40 | 1.0M | fp4 |
| Wafer | 112 tok/s | 0.45 s | $0.079 | $1.40 | 1.0M | fp4 |
| Decart | 111 tok/s | 1.06 s | $0.840 | $2.71 | 1.0M | fp4 |
| Wafer | 110 tok/s | 0.39 s | $0.179 | $1.40 | 1.0M | fp4 |
| Makora | 109 tok/s | 0.92 s | $0.140 | $4.40 | 1.0M | fp4 |
| ModelRun [by Modular] | 102 tok/s | 0.43 s | $2.10 | $6.60 | 1.0M | fp4 |
| Fireworks Spot | 101 tok/s | 0.68 s | $0.980 | $3.08 | 1.0M | – |
| Inceptron | 101 tok/s | 2.37 s | $0.600 | $2.80 | 1.0M | fp4 |
| Friendli | 97 tok/s | 0.68 s | $1.26 | $3.96 | 1.0M | – |
| Mistral | 95 tok/s | 0.67 s | $1.40 | $4.40 | 1.0M | nvfp4 |
| Modal | 92 tok/s | 0.43 s | $1.40 | $4.40 | 1.0M | – |
| CoreWeave | 91 tok/s | 0.94 s | $1.20 | $4.00 | 1.0M | fp8 |
| DeepInfra | 90 tok/s | 1.07 s | $0.563 | $2.50 | 1.0M | fp4 |
| Baidu Qianfan | 85 tok/s | 0.65 s | $0.599 | $1.88 | 1.0M | fp8 |
| Together | 84 tok/s | 1.42 s | $1.40 | $4.40 | 1.0M | – |
| Nebius Token Factory | 83 tok/s | 0.55 s | $1.40 | $4.40 | 1.0M | fp4 |
| Tenstorrent | 81 tok/s | 1.18 s | $0.600 | $2.50 | 1.0M | fp8 |
| Alibaba Cloud Int. | 80 tok/s | 1.07 s | $2.80 | $8.80 | 1M | – |
| Parasail | 78 tok/s | 0.86 s | $1.40 | $4.40 | 1.0M | fp8 |
| Sail Research | 75 tok/s | 1.29 s | $0.500 | $2.00 | 1.0M | fp8 |
| Amazon Bedrock | 75 tok/s | 5.36 s | $1.68 | $5.28 | 1.0M | – |
| DigitalOcean | 74 tok/s | 0.54 s | $0.910 | $2.86 | 1.0M | – |
| Sail Research | 73 tok/s | 0.66 s | $0.500 | $2.00 | 1.0M | fp8 |
| Fireworks | 72 tok/s | 2.18 s | $2.10 | $6.60 | 1.0M | – |
| Phala | 69 tok/s | 2.54 s | $0.840 | $2.64 | 1.0M | – |
| AkashML | 67 tok/s | 2.08 s | $0.190 | $4.40 | 1.0M | fp8 |
| DekaLLM | 63 tok/s | 1.36 s | $0.250 | $2.95 | 1.0M | – |
| NovitaAI | 57 tok/s | 1.44 s | $0.630 | $1.98 | 1.0M | fp8 |
| Z.ai | 57 tok/s | 2.24 s | $1.40 | $4.40 | 1.0M | fp8 |
| Reka AI | 53 tok/s | 0.91 s | $0.140 | $4.20 | 1.0M | – |
| Venice | 52 tok/s | 1.56 s | $1.40 | $4.40 | 1M | – |
| Baseten (fp4) | 50 tok/s | 5.11 s | $1.40 | $4.40 | 1.0M | fp4 |
| Baseten (fp4) | 48 tok/s | 3.71 s | $1.40 | $4.40 | 1.0M | fp4 |
| Relace | 47 tok/s | 3.23 s | $0.750 | $2.40 | 1.0M | – |
| SiliconFlow | 45 tok/s | 4.57 s | $0.700 | $2.20 | 1.0M | fp8 |
| Cloudflare | 45 tok/s | 1.81 s | $1.40 | $4.40 | 1.0M | – |
| Alibaba Cloud Int. | 39 tok/s | 1.50 s | $1.19 | $3.74 | 1M | – |
| AtlasCloud | 37 tok/s | 1.65 s | $1.40 | $4.40 | 1.0M | fp8 |
| Morph | 31 tok/s | 1.14 s | $0.439 | $4.40 | 1.0M | fp8 |
| GMICloud | 16 tok/s | 10 s | $0.700 | $2.20 | 1.0M | fp8 |
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 $0.260 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0019 | $0.0015 | 30.1 s |
| Summarize a 30-page report | 12,000 / 600 | $0.019 | $0.0092 | 33.6 s |
| Code edit | 6,000 / 1,500 | $0.015 | $0.0099 | 44.2 s |
| Agentic coding session | 60,000 / 4,000 | $0.102 | $0.050 | 1.2 min |
| Structured extraction | 2,000 / 200 | $0.0037 | $0.0020 | 28.9 s |
More from Zhipu AI
Compare GLM 5.3 with…
- GLM 5.3 vs Claude Opus 5.5
- GLM 5.3 vs Gemini 4 Argon
- GLM 5.3 vs Muse Wonder Full
- GLM 5.3 vs Claude Fable 5.1
- GLM 5.3 vs GPT-6 Astra
- GLM 5.3 vs GPT-6.1 Sol
- GLM 5.3 vs Claude Fable 5
- GLM 5.3 vs Claude Sonnet 5.5
- GLM 5.3 vs Claude Opus 5
- GLM 5.3 vs GPT-5.6 Sol
- GLM 5.3 vs Muse Wonder
- GLM 5.3 vs Muse Spark 1.3
- GLM 5.3 vs Kimi K3.1 Pro
- GLM 5.3 vs GPT-6 Sol
- GLM 5.3 vs GPT-5.5
- GLM 5.3 vs GPT-5.5 Pro
- GLM 5.3 vs GPT-5.4
- GLM 5.3 vs Kimi K3
- GLM 5.3 vs Gemini 3.8 Flash
- GLM 5.3 vs GPT-5.6 Terra
- GLM 5.3 vs Muse Spark
- GLM 5.3 vs Grok 4.6
- GLM 5.3 vs MiMo-V2.6-Pro
- GLM 5.3 vs DeepSeek V4.1 Flash 1017
- GLM 5.3 vs Grok 4.7
- GLM 5.3 vs GPT-5.3-Codex
- GLM 5.3 vs Gemini 3.7 Flash
- GLM 5.3 vs Qwen3.8 Max
- GLM 5.3 vs Claude Opus 4.7
- GLM 5.3 vs Claude Opus 4.8
- GLM 5.3 vs GPT-5.4 Pro
- GLM 5.3 vs Muse Spark 1.2
- GLM 5.3 vs Gemini 3.5 Flash
- GLM 5.3 vs Qwen3.8 2.4T A95B
- GLM 5.3 vs Claude Opus 4.6
- GLM 5.3 vs GLM 5.3 Flash
- GLM 5.3 vs Muse Spark 1.1
- GLM 5.3 vs GLM 5.2
- GLM 5.3 vs Qwen3.6 Max
See also
Questions people ask
- What are GLM 5.3's benchmark scores?
- As of 24 Oct 2026, GLM 5.3 (max effort) scores 63.5 ±3.2 on the BenchLeader Index, top 100, strongest in composite (80) and weakest in agents & tools (58). That rests on 30 independent benchmark results, and every score on this page links to its publisher.
- How does GLM 5.3 rank?
- GLM 5.3 (max effort) is #63 of 785 ranked configurations on 24 Oct 2026 (top 100). 136 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does GLM 5.3 cost per 1M tokens?
- As of 24 Oct 2026, GLM 5.3 costs $1.40 per 1M input tokens and $4.40 per 1M output tokens, $2.15 blended (three input tokens per output token).
- Is GLM 5.3 better than Gemini 3.7 Flash?
- As of 24 Oct 2026, Gemini 3.7 Flash (high) and GLM 5.3 (max) are neck and neck on quality (63.5 vs 63.5 on the BenchLeader Index, inside the ±4.7 margin of error), and Gemini 3.7 Flash (high) costs 1.4× less, which makes it the better buy for most workloads. The head-to-head page has the benchmark-by-benchmark detail.
- When was GLM 5.3 released?
- Zhipu AI released GLM 5.3 on 18 Aug 2026. BenchLeader has results for it at 4 reasoning-effort settings.
Data as of 24 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “GLM 5.3: benchmarks, pricing, speed and rank”, data as of 24 Oct 2026, https://www.benchleader.com/models/glm-5-3.