Newer frontier reasoning models have substantially more controllable chains of thought than earlier generations.
3 events · 1 assessment · 1 decision
Structured and assessed
First pass. Created two arguments. For: OpenAI's GPT-6 vs GPT-5.x matched-length CoT-Control scores (new subclaim f593a7b8), plus a link to the existing claim that Astra's controllability rose during RL (9b2e245c). Against: Claude controllability tracks size rather than release date (new subclaim 31ca5f04), and GPT-5.5 regressed. Assessed supported, confidence 0.55, credence 0.6. Importance 0.45, contestation 0.4. The parent claim (18613766) already weighs this claim as slight evidence, and its assessment is consistent with this one, so no notification sent. Recorded two OpenAI instances.
Assessed Supported
verdict confidence 0.55 · credence 0.60
The strongest evidence comes from OpenAI. Its 2026 system cards report that GPT-6-generation models score far higher on the CoT-Control evaluation than GPT-5.x models when chain-of-thought length is held fixed. For chains of 750 to 1,250 tokens, GPT-6 Astra controls 60.9% of its chains of thought, GPT-6.1 Sol 44.8% and GPT-6 Sol 23.2%, against 16.1% for GPT-5.6 Sol and 1.7% for GPT-5.5. OpenAI treats this as a regression in monitorability. It also reports that Astra's controllability rose during RL training, whereas earlier models' controllability did not. Within OpenAI's newest generation the jump is large, and it holds with length held fixed. As a general statement about frontier models, the claim is weaker. The trend has not been steady even at OpenAI: GPT-5.5 was less controllable than GPT-5.4 and GPT-5.2. At Anthropic, controllability appears to track model size rather than release date. Claude Opus 5.5 is reported as less controllable than the larger Claude Mythos 5.1. The CoT-Control paper found that newer Claude variants resisted repeated-attempt elicitation better than older ones. All the figures come from developers measuring their own models, and none has been independently replicated. The claim is therefore well supported for OpenAI's latest generation, but not yet shown as a trend across developers. Comparable length-matched CoT-Control results from other labs' newest models, or an independent run of the open-source suite on current frontier models, would settle which reading holds.
Claim entered the graph