Call GLM 4.6 Derestricted v5 through one API. 131K-token context, streaming and pay-per-token pricing. Try free in the Muapi playground. GLM 4.6 Derestricted v5 is a derestricted GLM 4.6 reasoning model with a 131K-token context window, modified to reduce refusal behavior.
GLM 4.6 Derestricted v5 is a derestricted GLM 4.6 reasoning model with a 131K-token context window, modified to reduce refusal behavior.
About this model
GLM 4.6 Derestricted v5 is a derestricted GLM 4.6 reasoning model with a 131K-token context window, modified to reduce refusal behavior. Muapi exposes it through a single REST endpoint with async polling and Server-Sent Events streaming, so you can drop it into chat apps, agents and workflows without hosting GPUs. It is best suited to long-context reasoning and general chat. Pricing is pay-per-token at $0.8/M input tokens and $3/M output tokens, with no subscription. See also GLM 5.3 Flash Abliterated, GLM 5.3 Abliterated. Browse the full lineup on the GLM Abliterated API page or the abliterated LLM API hub; related families: Gemma Abliterated API. 'Abliterated' and 'derestricted' describe modifications that reduce the model's refusal behavior. They do not guarantee unrestricted output, and you remain responsible for complying with applicable laws and the Muapi terms of use.
Cost analysis
| Provider | Cost | Notes |
|---|---|---|
| muapiapp | $0.8/M input tokens and $3/M output tokens | Pay-per-token, billed on actual usage with a small minimum reserved up front. |
| Fal.ai | Not available | GLM 4.6 Derestricted v5 is not offered as a hosted endpoint. |
| Replicate | Not available | GLM 4.6 Derestricted v5 is not offered as a hosted endpoint. |
Pay-per-token, billed on actual usage with a small minimum reserved up front.
GLM 4.6 Derestricted v5 is not offered as a hosted endpoint.
GLM 4.6 Derestricted v5 is not offered as a hosted endpoint.
** Competitor pricing is estimated based on similar model architectures and usage tiers.
Configuration schema
| Parameter | Type | Description | Default |
|---|---|---|---|
| Prompt | string | The user message or instruction for the model. | Summarize the trade-offs between a monolith and microservices in five bullet points. |
| System Prompt | string | Optional system-level instruction to guide model behavior. | You are a concise, direct assistant. |
| Max Tokens | integer | Optional maximum number of tokens to generate. | 8192 |
The user message or instruction for the model.
Summarize the trade-offs between a monolith and microservices in five bullet points.Optional system-level instruction to guide model behavior.
You are a concise, direct assistant.Optional maximum number of tokens to generate.
8192Developer documentation
Get an API key: Create a key in your Muapi dashboard.
Send a prompt: POST prompt (plus optional system_prompt and max_tokens, up to 32768) to /api/v1/glm-4-6-derestricted-v5.
curl -X POST https://api.muapi.ai/api/v1/glm-4-6-derestricted-v5 \
-H "x-api-key: $MUAPI_KEY" \
-H "Content-Type: application/json" \
-d '{"prompt": "Explain how attention works in transformers."}'
GET /api/v1/predictions/{request_id}/result until status is completed, or stream tokens live from /api/v1/glm-4-6-derestricted-v5/stream.curl -X POST https://api.muapi.ai/api/v1/glm-4-6-derestricted-v5/stream \
-H "x-api-key: $MUAPI_KEY" \
-H "Content-Type: application/json" \
-d '{"prompt": "Explain how attention works in transformers."}'
outputs. Usage is billed on actual token counts.Frequently asked
GLM 4.6 Derestricted v5 is a derestricted GLM 4.6 reasoning model with a 131K-token context window, modified to reduce refusal behavior.
Abliterated or derestricted models are modified to reduce refusal behavior. This lowers, but does not eliminate, refusals and does not guarantee unrestricted output.
Billing is per token: $0.8/M input tokens and $3/M output tokens. A small minimum is reserved when the request starts and reconciled to actual usage.
Yes. Stream tokens live from `/api/v1/glm-4-6-derestricted-v5/stream` using Server-Sent Events, or poll the async endpoint.
The context window is about 131K tokens. Prompt and output together must fit inside it; `max_tokens` is capped at 32768. The endpoint accepts text input only.
No. Abliterated models are less likely to refuse, but behavior varies by prompt and quality can differ from the base model. Evaluate it for your use case and follow applicable laws.