Today's price · per 1M tokens
Input
$1.40
per 1M tokens
Output
$4.40
per 1M tokens
Blended
$2.15
blended $/1M — 3:1 weighted input:output
Cached input
$0.260
19% of input — prompt caching
Price receipt
We read Z.AI's own pricing page on and found GLM-5.1 listed with a price on the same row — that read is where the number above comes from. We keep the page text we read; its content hash is aac79925f8.
Overview
GLM-5.1 model. $1.40/$4.40 per 1M; cached input $0.26.
Deploy this open model on rented GPUs
Open weights mean you can self-host instead of paying per-token API prices. These platforms let you serve it on demand — often cheaper at scale.
Partner links — we may earn a commission if you sign up, at no cost to you. We only list platforms we'd recommend regardless.
Capabilities
struck through = not supportedBenchmark performance
accuracy % · higher is betterSources: Z.AI model card (GPQA Diamond, SWE-Bench Verified, Humanity's Last Exam, ARC-AGI 2, AIME 2025, MMMLU, BFCL, HumanEval, MATH 500)
Every per-benchmark score we hold for this model is reported by its own vendor, so none is counted toward an accuracy average or any ranking. The percentile above is its standing across the independent composites below.
- GDPval-AA v2: 1124.7 ELO (Real-world work tasks, human baseline = 1000)
- Arena (arena.ai): 1465.1 ELO (Human preference)
- Arena WebDev: 1508.1 ELO (Human preference on building web apps)
- Epoch Capabilities Index: 150.7 (CI 148.7–152.8) (Epoch AI's index fitted across many benchmarks)
- FrontierMath Tiers 1–3: 36.8% (Research-level maths (funded by OpenAI, run by Epoch); best run across the effort settings Epoch tested)
Source: Artificial Analysis (artificialanalysis.ai)
- GDP.pdf: 8.4% (reasoning effort, measured 2026-11-22) — Tasks over long professional documents; share of tasks that pass every criterion
- AutomationBench-AA: 20.3% (reasoning effort, measured 2026-11-22) — Multi-step work across simulated business apps, without breaking the rules set
- Humanity's Last Exam: 30.1% (reasoning effort, measured 2026-11-22)
- GPQA Diamond: 86.8% (reasoning effort, measured 2026-11-22)
- AA-LCR (long-context reasoning): 73.7% (reasoning effort, measured 2026-11-22)
- CritPt (research-level physics): 4.6% (reasoning effort, measured 2026-11-22)
- IFBench (instruction following): 76.3% (reasoning effort, measured 2026-11-22)
- τ²-Bench Telecom: 97.7% (reasoning effort, measured 2026-11-22)
- Terminal-Bench Hard: 43.2% (reasoning effort, measured 2026-11-22)
- Terminal-Bench 2.1: 61.8% (reasoning effort, measured 2026-11-22)
- Terminal-Bench 4.0: 2% (reasoning effort, measured 2026-11-22)
- SciCode: 44.8% (reasoning effort, measured 2026-11-22)
- Ranks #67 of 207 comparably-measured models by percentile score, across 2 independent measurements
- Ranks #124 of 207 comparably-tested models by normalized performance per dollar
160 tokens/sec output
Throughput: Artificial Analysis via Vellum, recorded 2026-08-01 · Source: Artificial Analysis (artificialanalysis.ai)
Source: Artificial Analysis, LMSYS Chatbot Arena (UC Berkeley), Z.AI model card · updated · See full rankings →
Specifications
- Provider
- Z.AI
- Context window
- 128K tokens
- Modality
- text
- Parameters
- Proprietary
- Open source
- Yes — open weights available
- Released
- Status
- Current
- Last updated
- Tags
Availability verified: — listed on Z.AI's own page