Z.ai · Open weights
GLM 4.6 Original
GLM-4.6, Zhipu's flagship text model with 256K context window and advanced reasoning capabilities. Direct via Z-AI (Zhipu).
- Reasoning
- Tools
- Input
- Text
- Output
- Text
- Discovery tier
- Open weights
Pricing
USD per 1M tokens- Input
- $0.50
- Output
- $2.00
- Cache read
- $0.10
Reference prices as published by the model provider. API example
Specifications
- Context window
- 256K
- Max output
- 65.5K
- Average output (7 days)
- 1.3K tokens
- Longer than 81% of 468 models
Side by side
GLM 4.6 Original compared
Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.
| GLM 4.6 OriginalZ.aiThis model | DeepSeek V3.1 Terminus (Thinking)DeepSeekRemove | |
|---|---|---|
| Provider | Z.ai | DeepSeek |
| Discovery tier | Open weights | Open weights |
| Added | Dec 11, 2025 | Jan 1, 2024 |
| Limits | ||
| Context window | 256KBest | 128K |
| Max output | 65.54K | 65.54KBest |
| Average output | 1.3K tokens | 362 tokens |
| Pricing, USD per 1M tokens | ||
| Input | $0.50 | $0.25Best |
| Output | $2.00 | $0.70Best |
| Cache read | $0.10 | — |
| Performance | ||
| Throughput | — | 21.7 tok/s |
| Time to first token | — | 8.2 s |
| Request success (24h) | 99.4% | 99.4% |
| Benchmark indexes | ||
| Intelligence Index | 14.9Best | 14.8 |
| Coding Index | — | 43.5 |
| Math Index | 44.3 | 89.7Best |
| Capabilities | ||
| Reasoning | Yes | No |
| Tools | Yes | Yes |
| Structured output | No | Yes |
| Reasoning effort | — | — |
Public evaluations
Benchmarks
Independent evaluation results, with how each score ranks among the models in the Noumen catalog.
- GLM 4.6 Original
- DeepSeek V3.1 Terminus (Thinking)
Intelligence Index
- GLM 4.6 Original14.9
- DeepSeek V3.1 Terminus (Thinking)14.8
GLM 4.6 Original: Better than 37% of 326 scored models
Indexes
Coding Index
Composite of the coding evaluations
- GLM 4.6 Original: —
- DeepSeek V3.1 Terminus (Thinking): 43.5
Math Index
Composite of the math evaluations
- GLM 4.6 Original: 44.3
- DeepSeek V3.1 Terminus (Thinking): 89.7Top
Agentic work
AutomationBench-AA
Workflow automation, penalized for guardrail violations
- GLM 4.6 Original: —
- DeepSeek V3.1 Terminus (Thinking): 2.7%
AutomationBench-AA tasks completed
Workflows finished end to end without a guardrail violation
- GLM 4.6 Original: —
- DeepSeek V3.1 Terminus (Thinking): 0.2%
AA-Briefcasev1.1
Agentic knowledge work, rated head to head
- GLM 4.6 Original: —
- DeepSeek V3.1 Terminus (Thinking): 458 Elo
GDPval-AAv2.1
Economically valuable occupational tasks, rated head to head
- GLM 4.6 Original: —
- DeepSeek V3.1 Terminus (Thinking): 709 Elo
Document reasoning
GDP.pdf
Professional reasoning over PDF documents, all-pass rate
- GLM 4.6 Original: —
- DeepSeek V3.1 Terminus (Thinking): 5.4%
AA-LCRv1.1
Reasoning across long documents
- GLM 4.6 Original: 26.3%
- DeepSeek V3.1 Terminus (Thinking): 69.3%Top
Reasoning and knowledge
Humanity’s Last Exam
Expert-level questions across academic fields
- GLM 4.6 Original: 5.5%
- DeepSeek V3.1 Terminus (Thinking): 16.4%Top
GPQA Diamond
Graduate-level science questions
- GLM 4.6 Original: 63.2%
- DeepSeek V3.1 Terminus (Thinking): 79.2%Top
MMLU-Pro
Multi-subject knowledge and reasoning
- GLM 4.6 Original: 78.4%
- DeepSeek V3.1 Terminus (Thinking): 85.1%Top
IFBench
Following precise output instructions
- GLM 4.6 Original: 36.7%
- DeepSeek V3.1 Terminus (Thinking): 57.0%Top
Coding
Terminal-Benchv4.0
Practical software engineering in a terminal
- GLM 4.6 Original: —
- DeepSeek V3.1 Terminus (Thinking): 0.0%
SciCode
Scientific computing in Python
- GLM 4.6 Original: —
- DeepSeek V3.1 Terminus (Thinking): 38.0%
LiveCodeBench
Recent competitive programming problems
- GLM 4.6 Original: 56.1%
- DeepSeek V3.1 Terminus (Thinking): 79.8%Top
Math
AIME 2025
Competition mathematics, 2025 problems
- GLM 4.6 Original: 44.3%
- DeepSeek V3.1 Terminus (Thinking): 89.7%Top
Legacy evaluations
Terminal-Bench Hard
Earlier terminal task suite
- GLM 4.6 Original: 28.8%
- DeepSeek V3.1 Terminus (Thinking): 30.3%Top
τ²-Bench Telecom
Tool use in a customer-support simulation
- GLM 4.6 Original: 76.9%Top
- DeepSeek V3.1 Terminus (Thinking): 37.1%
AA-LCR (unversioned)
Long-context reasoning, earlier grading
- GLM 4.6 Original: 26.3%
- DeepSeek V3.1 Terminus (Thinking): 69.3%Top
Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis
Live routing data
Performance and reliability
Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.
- Prompt cache hits
- 24.4%
- About $0.40 per 1M input tokens with caching
- Request success (24h)
- 99.4%
- Reliable
Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.
- Healthy (98% or more)
- Degraded (90–98%)
- Down (under 90%)
- No requests
Developers
Use it through the API
Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.
curl https://www.noumen.xyz/api/v1/chat/completions \
-H "Authorization: Bearer $NOUMEN_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "z-ai/glm-4.6-original",
"messages": [{ "role": "user", "content": "Explain this stack trace." }]
}'import os
from openai import OpenAI
client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
model="z-ai/glm-4.6-original",
messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)Z.ai
Related models
Other Z.ai models, starting with the same family.