Uncensored Model Based on GLM-5.3 Released with 1M Context Window
Miles_Brundage · x · 2026-09-01
Twitter user @abliterationai released abliterated-model-large-v2, a model based on GLM-5.3. It ranks #3 on Terminal-Bench 4.0 (behind Opus 5 and Fable) and claims 2× the cyber exploitation capabilities of version 5.2.
Key features include:
- Hosted in the US
- FP8 precision
- 1 million context window
- Zero input/output prompt retention
The model is designed for offensive cyber operations, red teaming, and agent testing tasks that other models refuse to perform.
Related event: Abliterated GLM-5.3 Model Release Sparks Cybersecurity Policy Concerns(9 posts)→
More from Models
- Mocked benchmark shows DeepSeek's document chart grounding badly misaligned — burkov · 2026-10-18
- Dev complains Claude's habit of saying "not done" is his biggest pain point — jasonkneen · 2026-10-18
- Open-weight TYGER speech decision model makes 80ms turn-taking calls for full-duplex voice agents — solyarisoftware · 2026-10-18
- Paying Google AI subscriber blocked: custom-domain Workspace accounts can't access AI Studio — ChrisRackauckas · 2026-10-18
- Running Ling 3.1 Flash (560B MoE, 25B Active) Across 4 DGX Sparks via llama.cpp RPC — BerryBlush_Rogers691 · 2026-10-18
- Users press OpenAI on long-delayed Adult Mode a year after it was promised — ___Patrice___ · 2026-10-18