Chain-of-thought extraction is a technique used to distill proprietary large language models

Overview Debates Related claims Topics

OpenAI's GPT-6.1 Terra model operated three times faster than both Sol and Astra on Agent's Last Exam

OpenAI's GPT-6.1 Terra model completed benchmark tasks on Agent's Last Exam at half the cost of Sol and one-tenth the cost of Astra

OpenAI made GPT-6.1 Terra available at 50% of the cost of GPT-5.6 Terra

OpenAI released the mid-sized artificial intelligence model GPT-6.1 Terra

Anthropic launched Claude Opus 5.5 to improve text clarity following feedback that Claude Opus 5 produced convoluted writing

Anthropic implemented preserved thinking blocks across conversation turns to protect the Anthropic API from model distillation

The research paper titled 'Stealing Reasoning Traces from Proprietary LLMs' demonstrated techniques for extracting reasoning traces from closed models

Massachusetts Institute of Technology and Stanford University interns at Alpha School worked 80 hours per week conducting research to build an educational large language model

Claude Sonnet 5.5 achieved strong performance on Leash Bio's chemical exploration benchmark

Claude Sonnet 5 exhibited high operational costs and low performance on Leash Bio's chemical exploration benchmark

Solve 6.1 demonstrated high performance on Leash Bio's chemical exploration benchmark

Ramp data showed Claude Opus 5.5 captured 35.1% of token share and 34.4% of business spend across tracked models

Moonshot released Kimi K3.1 Pro as the first model update to the Kimi K3 family

Artificial Analysis determined that Kimi K3.1 consumes 14% more tokens per task than Kimi K3

Moonshot priced Kimi K3.1 at the same per-token rate as Sonnet 5.5 and GPT6 SOL

Moonshot instructed influencers to promote Kimi K3.1 as matching Opus 5.5 performance at one-fifth the cost

Moonshot's open-weight Kimi K3 model had 2.8 trillion parameters upon its July release

In independent evaluations by Artificial Analysis, Kimi K3.1 scored 55 on the Intelligence Index, ranking fourth behind Opus 5.5, Sonnet 5.5, and Muse

Astra 6 achieved the highest performance among frontier language models evaluated on Leash Bio's chemical exploration benchmark

On the Artificial Analysis Intelligence Index, Kimi K3.1 scored two points higher than Fable 5.1, GPT6 Astra, and Gemini 4 Argon