Breaking

GPT-6.1 Sol and Claude Sonnet 5.5 intensify the push for lower-cost AI agents

September 30, 2026 at 15:41 EDT

Foundation Models
AI Agents
Software Dev & Coding

OpenAI and Anthropic have released lower-cost workhorse models aimed at agentic coding, computer use and professional workflows, setting up a fresh comparison with premium systems such as GPT-6 Astra, Claude Opus 5.5 and Claude Fable 5.1. The core question for developers is no longer only which model has the highest benchmark ceiling, but which one can finish real tasks at an acceptable quality and cost.

Sep 30, 2026 · OpenAI / Anthropic

The $2 agent tier gets serious

GPT-6.1 Sol and Claude Sonnet 5.5 bring high-end coding, computer-use and work-agent capabilities closer to production budgets, while premium models still hold the hardest-work ceiling.

$2/$10

GPT-6.1 Sol and Sonnet 5.5 standard input/output per 1M tokens

$0.10

GPT-6.1 Sol cached input per 1M tokens

30%+

Sonnet 5.5 speed gain versus Sonnet 5

1.05M

GPT-6.1 Sol context window in tokens

Standard token price: Sol versus Astra

GPT-6.1 Sol

GPT-6 Astra

Astra's standard input and output token prices are five times Sol's.

What buyers like

Sol is attractive for repeated-context agents, reviews and audits. Sonnet 5.5 is attractive for fast iteration, everyday coding and polished work products.

What still limits it

Premium models remain better suited to the hardest open-ended work, and higher effort settings can narrow or erase apparent savings.

The new evaluation loop

Measure task quality

Track real tokens

Route by task risk

The practical question is cost per completed task, not model rank alone.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.