Abliterated Model Large V2, and what to use instead

abliterated-model-large-v2 is abliteration.ai's own model: an abliterated, retuned build of the open-weight GLM-5.3 with a 1M token context. Wild West API does not serve it. The nearest model it does serve is GLM 5.3 Xploded, an uncensored build of the same GLM 5.3 base, also with 1M of context and tool calling.

What abliterated-model-large-v2 is

abliteration.ai sells three models of its own through an OpenAI- and Anthropic-compatible API. Its documentation describes Large V2 as derived from the open-weight GLM-5.3, further abliterated and fine-tuned by Abliteration AI. It is text only, reads up to 1,000,000 tokens of context, streams, calls tools, returns structured output, and reasons before answering in one of three modes: low, high or max. A request to switch reasoning off on Large V2 is mapped to hidden low reasoning instead.

"Large" is the GLM base, a frontier-scale open model; "v2" marks the move from GLM-5.2, the base of the previous Large, to GLM-5.3. Abliterated means the refusal direction was removed from the weights, and the retune after it is what wins back the capability that edit costs.

abliteration.ai's models against the nearest ones here

abliteration.ai figures are from their models and pricing pages, checked 2 October 2026, standard rates before any plan discount. Wild West API prices are live at build time.

Which to use

If you want abliterated weights specifically, on GLM-5.3, abliteration.ai is the place that has them, and it adds things Wild West API does not: web search, a policy gateway with audit export, and three reasoning modes. The side-by-side comparison is written to be fair to it.

If what you want is the GLM 5.3 base with the refusals gone, behind a key that cannot spend past a cap you set, GLM 5.3 Xploded is that: per-token pricing with no plan, and a hard per-key ceiling enforced before the request leaves. If abliterated is the requirement, Outlaw 1 Xploded is the abliterated model on the line.

Calling GLM 5.3 Xploded

Any OpenAI client works by changing the base URL. A request written for abliteration.ai needs nothing else changed except the model id.

curl https://wildwestapi.com/v1/chat/completions \
  -H "Authorization: Bearer sk-ww-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm-5.3-xploded",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

From Python

# pip install openai
from openai import OpenAI

client = OpenAI(
    base_url="https://wildwestapi.com/v1",
    api_key="sk-ww-...",
)

resp = client.chat.completions.create(
    model="glm-5.3-xploded",
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)

FAQ

Read further