Breaking

OpenAI Launches GPT-6.1 Sol With Cheaper Cached Input and Stronger Agentic Coding

September 29, 2026 at 13:55 EDT

Foundation Models
AI Agents
Software Dev & Coding

OpenAI introduced GPT-6.1 Sol on September 29, 2026, positioning it as a lower-cost upgrade to GPT-6 Sol for coding agents, computer-use workflows, and professional work. The model is available now in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users, and through the OpenAI API as gpt-6.1-sol; it is not yet available in Chat.

Sep. 29, 2026 · OpenAI

GPT-6.1 Sol shifts the agent-cost curve

The new Sol model is built for long-running coding and computer-use agents, with near-Astra performance claims and a sharply cheaper cache-read price.

95%

cached-input discount versus standard input

1.05M

token context window

128K

maximum output tokens

planned Ultrafast token generation boost

Input price becomes a cache story

Same unit: dollars per 1 million input tokens. Each block represents $0.10.

$2.00

standard input

$0.10

cached input

A reused context read is priced at 1 block instead of 20.

Why it matters for agents

Reuse context

large codebases and documents

→

Run tools

computer use, shell, files

→

Iterate longer

refactors and investigations

Upside

Lower cached-input pricing, stronger coding benchmarks, and a 1.05M-token window make it a better fit for persistent agents.

Watchpoints

It is not yet in Chat, Astra still leads some hardest tasks, and developers are expected to compare quality on their own workloads.

Continue reading

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