Skip to content
Trending storyDeveloping

Epoch AI measures faster long-prompt handling in GPT-6.1 Sol, suggesting a possible change

2 articles1 sourceLast article 6h ago

Overview

AI overview

Epoch AI reports that GPT-6.1 Sol handles long prompts faster than GPT-6 Sol, and says its measurements suggest something has changed in how the model processes long contexts.

OpenAI has halved the cached-input price relative to GPT-6 Sol, which Epoch cites alongside the speed results. Epoch explicitly says this is not conclusive evidence of a new architecture, so a fundamental design change is only a possibility rather than an established finding.

The speed claim comes from Epoch's own latency measurements, which it describes as a follow-up to an earlier report on latency scaling across frontier models. The evidence available here is limited to Epoch's statements on X and a link to that appendix; the underlying measurement details are not reproduced in these reports.

AIWritten by AI from the articles below · overview updated Oct 8, 9:04 PM ET

Check the sources:

Latest developmentOct 8, 5:19 PM ET
This isn’t conclusive evidence of a new architecture, but it suggests something has changed in how GPT-6.1 Sol handles long contexts. This is a follow-up to our earlier report on latency scaling in frontier models: https://epoch.ai/publications/long-context-latency-scaling-gpt-vs-claude#appendix-h-gpt-6-sol-and-gpt-61-sol

Developments

2 developments

  1. Oct 8, 5:19 PM ET · 1 article
    Epoch AI reports latency change suggesting altered long-context handling in GPT-6.1 Sol
    Epoch AI: This isn’t conclusive evidence of a new architecture, but it suggests something has changed in how GPT-6.1 Sol handles long contexts. This is a follow-up to our earlier report on latency scaling in frontier models: https://epoch.ai/publications/long-context-latency-scaling-gpt-vs-claude#appendix-h-gpt-6-sol-and-gpt-61-sol
  2. Oct 8, 5:19 PM ET · 1 article
    Epoch AI reports GPT-6.1 Sol halves cached-input price and handles long prompts faster than GPT-6 Sol
    Epoch AI: A new architecture for GPT-6.1 Sol? OpenAI halved its cached-input price compared with GPT-6 Sol. Our measurements show it also handles long prompts faster.

Article timeline

Follow the coverage from different perspectives. Times are ET.

Oct 8
  1. Epoch AI
    This isn’t conclusive evidence of a new architecture, but it suggests something has changed in how GPT-6.1 Sol handles long contexts. This is a follow-up to our earlier report on latency scaling in frontier models: https://epoch.ai/publications/long-context-latency-scaling-gpt-vs-claude#appendix-h-gpt-6-sol-and-gpt-61-sol

    This isn’t conclusive evidence of a new architecture, but it suggests something has changed in how GPT-6.1 Sol handles long contexts. This is a follow-up to our earlier report on latency scaling in frontier models: https://epoch.ai/publications/long-context-latency-scaling-gpt-vs-claude#appendix-h-gpt-6-sol-and-gpt-61-sol

  2. Epoch AI
    A new architecture for GPT-6.1 Sol? OpenAI halved its cached-input price compared with GPT-6 Sol. Our measurements show it also handles long prompts faster.

    A new architecture for GPT-6.1 Sol? OpenAI halved its cached-input price compared with GPT-6 Sol. Our measurements show it also handles long prompts faster.

Heat trend

Not enough continuous observations to show a trend yet.