abliteration · Frontier
Abliterated Model Large
Abliteration.ai's unrestricted large text reasoning model is derived from GLM-5.2 and weight-modified to reduce refusals compared with the base model. It supports native tool calling, structured output, automatic prompt caching, and a one-million-token context window.
- Reasoning
- Tools
- Structured output
- Input
- Text
- Output
- Text
- Reasoning effort
- none · minimal · low · medium · high · xhigh · max
- Discovery tier
- Frontier
Pricing
USD per 1M tokens- Input
- $3.00
- Output
- $5.00
- Cache read
- $0.30
Reference prices as published by the model provider. API example
Specifications
- Context window
- 1M
- Max output
- 1M
- Average output (7 days)
- 467 tokens
- Longer than 47% of 473 models
Side by side
Abliterated Model Large compared
Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.
| Abliterated Model LargeabliterationThis model | Claude Sonnet 4.6AnthropicRemove | |
|---|---|---|
| Provider | abliteration | Anthropic |
| Discovery tier | Frontier | Frontier |
| Added | Aug 27, 2026 | Feb 17, 2026 |
| Limits | ||
| Context window | 1M | 1M |
| Max output | 999.99KBest | 128K |
| Average output | 467 tokens | 401 tokens |
| Pricing, USD per 1M tokens | ||
| Input | $3.00 | $3.00 |
| Output | $5.00Best | $15.00 |
| Cache read | $0.30 | $0.30 |
| Performance | ||
| Throughput | 131.7 tok/sBest | 78.6 tok/s |
| Time to first token | 3.7 s | 1.5 sBest |
| Request success (24h) | 98.9% | 99.5%Best |
| Benchmark indexes | ||
| Intelligence Index | — | 24.7 |
| Capabilities | ||
| Reasoning | Yes | Yes |
| Tools | Yes | Yes |
| Vision | No | Yes |
| Structured output | Yes | Yes |
| PDF input | No | Yes |
| Reasoning effort | none, minimal, low, medium, high, xhigh, max | low, medium, high, xhigh, max |
Public evaluations
Benchmarks
Independent evaluation results, with how each score ranks among the models in the Noumen catalog.
- Abliterated Model Large
- Claude Sonnet 4.6
Intelligence Index
- Abliterated Model Large—
- Claude Sonnet 4.624.7
Document reasoning
AA-LCRv1.1
Reasoning across long documents
- Abliterated Model Large: —
- Claude Sonnet 4.6: 68.3%
Reasoning and knowledge
Humanity’s Last Exam
Expert-level questions across academic fields
- Abliterated Model Large: —
- Claude Sonnet 4.6: 13.3%
GPQA Diamond
Graduate-level science questions
- Abliterated Model Large: —
- Claude Sonnet 4.6: 79.9%
IFBench
Following precise output instructions
- Abliterated Model Large: —
- Claude Sonnet 4.6: 41.2%
Legacy evaluations
Terminal-Bench Hard
Earlier terminal task suite
- Abliterated Model Large: —
- Claude Sonnet 4.6: 46.2%
τ²-Bench Telecom
Tool use in a customer-support simulation
- Abliterated Model Large: —
- Claude Sonnet 4.6: 79.5%
AA-LCR (unversioned)
Long-context reasoning, earlier grading
- Abliterated Model Large: —
- Claude Sonnet 4.6: 68.3%
Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis
- Abliterated Model Large
- Claude Sonnet 4.6
| Leaderboard | Abliterated Model Large | Claude Sonnet 4.6 |
|---|---|---|
| Coding | Not in top 20 | #18 · 1528.2 |
| Longer Query | Not in top 20 | #18 · 1492.3 |
| Industry Life And Physical And Social Science | Not in top 20 | #18 · 1499.6 |
| Vision Style Control · English | Not in top 20 | #9 · 1282.8 |
| Document Overall | Not in top 20 | #8 · 1482.2 |
| Document Style Control Overall | Not in top 20 | #10 · 1477.7 |
Ranks among Noumen catalog models on LMArena leaderboards. Only top-20 placements are listed.LMArena
- Abliterated Model Large
- Claude Sonnet 4.6
| Leaderboard | Abliterated Model Large | Claude Sonnet 4.6 |
|---|---|---|
| SWE-bench Verified | Not in top 20 | #8 · 75.2% |
Ranks among Noumen catalog models on Epoch AI leaderboards. Only top-20 placements are listed.Epoch AI
- Abliterated Model Large
- Claude Sonnet 4.6
| Leaderboard | Abliterated Model Large | Claude Sonnet 4.6 |
|---|---|---|
| EQ-Bench 4 | Not in top 20 | #19 · 1200 |
| Longform Writing | Not in top 20 | #16 · 79.9 |
| EQ-Bench 3 | Not in top 20 | #5 · 1710 |
Ranks among Noumen catalog models on EQ-Bench leaderboards. Only top-20 placements are listed.EQ-Bench
- Abliterated Model Large
- Claude Sonnet 4.6
| Leaderboard | Abliterated Model Large | Claude Sonnet 4.6 |
|---|---|---|
| ASCII Art | Not in top 20 | #17 · 1233 |
| Agentic Game Dev | Not in top 20 | #14 · 1140 |
| Full Stack | Not in top 20 | #14 · 1213 |
| Native Apps | Not in top 20 | #10 · 1198 |
| Mobile Apps | Not in top 20 | #13 · 1211 |
Ranks among Noumen catalog models on Design Arena leaderboards. Only top-20 placements are listed.Design Arena
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.
- Throughput
- 131.7 tok/s
- Faster than 70% of 256 models
- Time to first token
- 3.7 s
- Quicker than 43% of 256 models
- Prompt cache hits
- 97.3%
- About $0.37 per 1M input tokens with caching
- Request success (24h)
- 98.9%
- 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": "abliteration-ai/abliterated-model-large",
"reasoning_effort": "medium",
"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="abliteration-ai/abliterated-model-large",
messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)abliteration
Related models
Other abliteration models, starting with the same family.