Short answer: Z.ai’s documentation and pricing page list GLM-5, a text-only model with a 200K-token context window, up to 128K output tokens, and features including function calling, streaming, context caching, and structured output. Z.ai identifies GLM-5.3 as its latest flagship. GLM-5 may be worth testing when its capabilities and listed price fit your workload; confirm account access and API behavior before building a production integration.
What GLM-5 is—and what it isn’t
Z.ai describes GLM-5 as a model for agentic engineering, complex systems work, and long-range agent tasks. Its model overview lists text input and output, a 200K context length, 128K maximum output tokens, and features such as thinking modes, streaming, function calling, context caching, and structured output. These are vendor-listed capabilities, not a guarantee of quality or reliability for a particular application.
Keep the model names distinct. GLM-5’s official overview lists text-only input. GLM-5V-Turbo is separately documented as a multimodal coding model; do not attribute its image or video capabilities to GLM-5. Z.ai describes GLM-5.3 as its latest flagship and lists a 1M-token context window with 128K maximum output. GLM-5 remains listed in Z.ai’s model documentation and pricing, but confirm that it is enabled for your account and intended route.
Is GLM-5 a fit for your workload?
Consider testing GLM-5 for text-based coding assistance, long-document work, structured extraction, or multi-step tasks involving tools. Its listed context length and features make it a candidate for those workflows; they do not establish that it will outperform another model on your codebase or prompts.
Compare GLM-5.3 if a larger context window or Z.ai’s latest flagship matters to your use case. Its documented context window is 1M tokens, compared with GLM-5’s 200K. The pricing page lists higher standard input and output rates for GLM-5.3; whether that trade-off is worthwhile depends on your actual task and token usage.
If your task needs image or video input, evaluate a model documented for those modalities instead. GLM-5’s overview lists text input; GLM-5V-Turbo is a separate product with its own specifications and API details.
Do not choose from benchmark headlines alone. Compare GLM-5 with your current or fallback model on representative tasks. Track task success, output-format validity, latency, token use, and human correction time. The cited model overview does not provide a comparable benchmark methodology or production SLA.
Z.ai API pricing and access
As checked on November 10, 2026, Z.ai’s pricing documentation lists prices in USD per 1 million tokens:
- GLM-5 input: $1.00
- GLM-5 cached input: $0.20
- GLM-5 cached-input storage: listed as limited-time free
- GLM-5 output: $3.20
- GLM-5.3 input: $1.40
- GLM-5.3 output: $4.40
These are listed token-based API prices, not coding-plan subscription prices or third-party aggregator rates. Check applicable billing terms and account access before purchase or rollout.
Worked estimate: At 10 million standard, uncached input tokens and 2 million output tokens, the listed GLM-5 rates yield $16.40: (10 × $1.00) + (2 × $3.20). At the listed GLM-5.3 rates, the same volumes yield $22.80. This arithmetic excludes cached-input discounts, taxes, tools, and other charges.
Z.ai’s Chat Completions API reference documents the endpoint https://api.z.ai/api/paas/v4/chat/completions and Bearer-token authentication. The GLM-5 overview and pricing page list the model name glm-5, but the documentation excerpts available for this guide do not provide a model-specific quick-start request confirming that identifier and every field below as a supported GLM-5 API contract. Treat this as an illustrative request pattern, not a verified, tested GLM-5 call; confirm it against the current API documentation and your account before relying on it.
curl --request POST \\
--url https://api.z.ai/api/paas/v4/chat/completions \\
--header "Authorization: Bearer $ZAI_API_KEY" \\
--header "Content-Type: application/json" \\
--data '{
"model": "glm-5",
"messages": [
{"role": "user", "content": "Summarize this incident report in five bullet points."}
]
}'
Keep the API key in a secrets manager or environment variable, not in source code. Confirm the accepted model identifier, request fields, account access, and applicable rate limits with Z.ai before production use.
Production-fit pilot: a go/no-go checklist
Start with one bounded workflow and a fallback model. For example, test whether GLM-5 can classify support tickets and return a fixed JSON object. Use representative test cases and define acceptance criteria before running the evaluation:
- Task quality: Does the output meet a reviewer’s rubric?
- Output validity: Is every response valid JSON with the required fields and types?
- Operational performance: Measure end-to-end latency, timeouts, and retry frequency under expected traffic.
- Total cost: Record input and output token counts, plus human review and correction time.
- Failure and fallback: Decide what to do when a request fails, returns unusable output, or exceeds your latency threshold.
- Provider terms: Confirm applicable rate limits, data-handling terms, and any contractual or service commitments for your account. The model overview and pricing page do not establish those guarantees.
Promote GLM-5 only if it meets your quality and format thresholds and its measured latency and total cost fit your requirements. If its 200K context is insufficient, or GLM-5.3’s larger context better fits the job, test the better-matched option.
Bottom line
GLM-5 is a candidate to evaluate, not an automatic production choice. Z.ai lists a text-only model with a 200K context window and API rates of $1.00 per million input tokens and $3.20 per million output tokens. For a new deployment, compare it with GLM-5.3 and your current model on representative tasks. Verify the exact API request contract, account limits, data terms, and failure behavior before launch.
Sources
GLM-5 - Overview - Z.AI Developer Documentation
GLM-5.3 - Overview - Z.AI Developer Documentation
Pricing - Z.AI Developer Documentation