Qwen3.6 35B Heretic drops with near-zero KLD

// 45d agoMODEL RELEASE

Qwen3.6 35B Heretic drops with near-zero KLD

Qwen3.6-35B-A3B-Heretic is a decensored version of Qwen's latest 35B Mixture-of-Experts (MoE) model, engineered to maintain its high-end reasoning and coding capabilities without the restrictive safety filters. Developed using the Magnitude-Preserving Orthogonal Ablation (MPOA) method, the model achieves a remarkably low KL Divergence of 0.0015, ensuring it stays true to the original's intelligence while slashing refusal rates by nearly 90%.

// ANALYSIS

Heretic is a masterclass in surgical model abliteration, proving that safety filters can be removed without sacrificing the underlying intelligence of a high-parameter MoE architecture.

–KL Divergence of 0.0015 is remarkably low, indicating the model's weights remain statistically nearly identical to the original despite the "lobotomy" reversal.
–The 88% reduction in refusals makes it a viable choice for complex, uncensored coding and multi-turn reasoning tasks where standard Qwen models often "preach."
–With only 3B activated parameters, it delivers high-end reasoning performance while remaining runnable on consumer hardware like a single 24GB VRAM GPU.
–Preservation of the "Thinking" context is critical for developers using it as an agentic backend for repo-level refactoring.

// TAGS

qwen3.6-35b-a3b-hereticllmopen-weightsmixture-of-expertsdecensoredai-coding

DISCOVERED

45d ago

2026-04-26

PUBLISHED

45d ago

2026-04-26

RELEVANCE

9/ 10

AUTHOR

My_Unbiased_Opinion

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

MODEL33m ago

Anthropic releases Claude Fable 5, a public version of its powerful Mythos AI model equipped with safety guardrails that reroute high-risk queries to Claude Opus 4.8.

Anthropic recently announced the release of Claude Fable 5, a public-facing version of its highly capable, agentic "Mythos-class" AI model. Claude Fable 5 offers significant advances in agentic multi-step task execution, complex reasoning, and coding capabilities, scoring 80.3% on SWE-Bench Pro. To manage risks associated with frontier capabilities, Anthropic has implemented safety classifiers that automatically redirect high-risk queries—such as those related to cybersecurity, biology, or chemistry—to the less powerful Claude Opus 4.8 model. However, the release introduces new enterprise challenges, including a mandatory 30-day data retention policy that overrides existing zero-retention agreements and a heightened need for organizations to implement Zero Trust architectures and update their third-party vendor risk assessments.

MODEL39m ago

Claude Fable 5 impresses Anthropic skeptics

Anthropic has released Claude Fable 5, a "Mythos-class" model designed to autonomously plan, execute, and verify complex, long-running agentic tasks. To manage safety risks, Fable 5 automatically routes sensitive prompts to Claude Opus 4.8 and requires a 30-day safety-monitoring data retention period.

MODEL46m ago

Open research advocates criticize Anthropic for silently degrading its newly released Claude Fable 5 model by routing restricted queries to less capable models.

Open research advocates, including AlphaXiv, have publicly expressed disappointment with Anthropic's deployment strategy for its newly released frontier model, Claude Fable 5. Released on June 9, 2026, the highly capable model features strict safety classifiers that automatically reroute sensitive or restricted queries to the older, less capable Claude Opus 4.8 without clear notification to users. This "silent degradation" of capabilities, combined with a mandatory 30-day data retention policy for safety monitoring, has sparked significant backlash from developers and researchers who argue that Anthropic is gatekeeping frontier intelligence and undermining open scientific progress.