How to Use GLM-5 in goose
GLM-5 is built for complex software engineering, including backend architecture, debugging, and tasks with several stages. Compared with GLM-5 Turbo's focus on persistent tool workflows, GLM-5 targets broader systems engineering.
Prompts to try with GLM-5 in goose
Copy an example into goose, or ask in your own words.
Use GLM-5 with thinking enabled to investigate the failing tests in this project. Reproduce the failure, identify the cause, make the smallest useful patch, and report which checks passed.
Use GLM-5 to review this project's backend architecture before we add organization accounts. Map the affected components, identify migration risks, and propose steps with acceptance criteria.
Use GLM-5 with temperature 0.7 and top-p 0.95 to investigate duplicate records in this project. Trace the write path, reproduce the problem, and verify a focused fix.
Ask GLM-5 through Spiral to review the diff on this branch and list the three riskiest changes. Save the result in ./assets.
Get a second opinion from GLM-5 on this function, then tell me what the call cost.
How to set up GLM-5 in goose
Pick the way that suits you. Each takes about two minutes.
Use Spiral's MCP server
- Add the extension
In goose Desktop, open the sidebar, click Extensions, then Add custom extension. Enter:
- Sign in
When goose first connects to Spiral, your browser opens. Sign in to Spiral, pick the organization and click Authorize. No API key. If the browser does not open, goose prints the link to open yourself.
Use the Spiral skill
- Send goose this message
Your agent reads Spiral's skill and sets itself up.
set up https://spiral.ai/SKILL.md
- Sign in or paste a key
Approve Spiral in the browser, or give the agent a key from the console.
Use the Spiral CLI
- Install and sign in
Installs the Spiral CLI, signs you in and adds its skill to your coding agents.
npx -y spiral-cli@latest init
Call the Spiral API
- Create an API key
In the console, under API keys.
- Run GLM-5
One request; the result comes back by polling or a webhook.
curl https://api.spiral.ai/v1/runs \ -H "Authorization: Bearer $SPIRAL_API_KEY" \ -H "Content-Type: application/json" \ -H "Idempotency-Key: $(uuidgen)" \ -d '{ "provider": "openrouter", "model": "z-ai/glm-5", "input": { "messages": [ { "role": "system" } ] } }'
Then paste one of the prompts above: goose finds GLM-5 on Spiral, shows the price and runs it.
GLM-5 providers and prices
Spiral runs GLM-5 through OpenRouter. Open a card for today's price, with every fee included.
Up to $0.017086 per run at the default inputs. $1.30 per 1M input tokens, $4.16 per 1M output tokens.
Settings you can ask for
Ask goose for any of these in your prompt.
| Setting | Allowed values | Default | Description |
|---|---|---|---|
| Seed | a whole number | ||
| Top k | a whole number from 0 | ||
| Top p | 0 to 1 | ||
| Logprobs | on or off | ||
| Max tokens | 1 to 131072 | The most output tokens to generate. | |
| Temperature | 0 to 2 | ||
| Top logprobs | 0 to 20 | ||
| Presence penalty | -2 to 2 | ||
| Frequency penalty | -2 to 2 | ||
| Repetition penalty | 0 to 2 |
Every API field of Z.ai: GLM 5 is on its tool page: open its card above.
What can you do with GLM-5 in goose?
Turn a project requirement or bug report into code changes with checks that demonstrate the result. GLM-5 can help trace dependencies and plan work across several files.
Reproduce and fix a backend bug
Give the agent a failing request, relevant logs, and expected behavior. Ask it to trace the cause, make a focused patch, and demonstrate the fix.
Plan a backend refactor
Map dependencies, identify behavior that must stay consistent, and divide the change into reviewable steps with acceptance checks.
Implement a feature from requirements
Turn a feature brief into an implementation plan, code changes, and a record of completed checks. State the expected user behavior before work begins.
Research a technical decision
Have the agent retrieve documentation, compare options against stated requirements, and produce a recommendation with supporting links and unresolved questions.
Extract records from text
Turn pasted notices, support requests, or meeting notes into consistently structured records. Define the expected structure and a rule for missing information.
Prepare a recurring operations brief
Combine recent tool results into a short report covering changes, unresolved problems, and next actions. Carry a compact summary between runs.
Investigate a data discrepancy
Ask the agent to compare reported totals, inspect relevant records through available tools, and explain which checks support its conclusion.
More on GLM-5 and goose
Overview
GLM-5 in other agents
Questions about GLM-5 in goose
Can goose use GLM-5?
Yes. Add Spiral to goose as an MCP server, and goose can find and run GLM-5 like any built-in tool.
How much does GLM-5 cost in goose?
Up to $0.017086 per run at the default inputs. $1.30 per 1M input tokens, $4.16 per 1M output tokens. You pay from your Spiral balance as you go, with no subscription, and a failed run is not charged. Your goose plan is separate.
Do I need an OpenRouter account to use GLM-5 in goose?
No. Spiral runs GLM-5 on its own OpenRouter account. You sign in to Spiral once and never paste a key.
How long does it take to set up GLM-5 in goose?
About two minutes: add the extension, then sign in.
Does goose wait while GLM-5 runs?
GLM-5 runs in the background. goose starts the run, checks on it and gives you the text, json when it's ready.
What is GLM-5 best suited for?
Use GLM-5 for software engineering and tasks that require planning, tool use, and several rounds of checking. GLM-5 emphasizes complex systems work; GLM-5 Turbo emphasizes persistent workflows with many actions.
How much text can GLM-5 handle?
Z.ai lists a 200K context window and a maximum output of 128K tokens for both GLM-5 and GLM-5 Turbo. Those figures describe capacity, so choose an answer length that fits the task.
When should I enable thinking?
Enable thinking for debugging, complex planning, and decisions with several constraints. Ask for a concise conclusion and verification evidence so the final answer stays useful.
How do I get useful JSON responses?
Specify the exact structure, required information, and how to handle missing values. Include a small example so the requested result is unambiguous.
How should I handle long research conversations?
Keep the current question, constraints, and relevant evidence in a compact working summary. The GLM-5 technical report found that accuracy can decline in extremely long contexts, so avoid carrying every old tool result forward.
Start using GLM-5 in goose
One Spiral account for every tool your agent needs. No subscription; you pay per run.
Paste this into goose:
set up https://spiral.ai/SKILL.md


