OpenAI Unveils GPT-6.1 Sol at DevDay With New Codex and ChatGPT Tools

OpenAI introduced GPT-6.1 Sol, an upgrade to its GPT-6 Sol model, at its DevDay 2026 event on September 29, 2026, releasing the model in the API at $2 per million input tokens and $10 per million output tokens. The release was part of more than 20 announcements detailed in the company’s DevDay 2026 recap, spanning Codex developer tools, the API, ChatGPT collaboration features, plugins, and subscription plans.

According to the GPT-6.1 Sol announcement, the model nearly matches GPT-6 Astra’s intelligence on agentic coding, computer use, and professional work at one-fifth of Astra’s standard input and output token prices. OpenAI said cached input costs $0.10 per million tokens, a price it describes as 95% below standard input pricing and 50% below GPT-6 Sol’s cached input price.

GPT-6.1 Sol is available starting September 29, 2026 to all Plus, Pro, Business, Enterprise, and Edu users in ChatGPT Work and Codex, and developers can access it through the OpenAI API as gpt-6.1-sol. It is not yet available in Chat. OpenAI said a GPT-6.1 Sol Ultrafast option, with up to eight times faster token generation than the model’s standard speed in Codex, will follow in the coming days. The company said the announcements open ChatGPT as a shared surface where humans and agents can collaborate and where developers can launch native experiences to what it describes as a collective 1.2 billion weekly users.

GPT-6.1 Sol Performance and Safety Results

On DeepSWE v1.1, a benchmark for complex software-engineering tasks in real codebases, OpenAI reports that GPT-6.1 Sol matches GPT-6 Astra at roughly one-fifth of the cost while exceeding GPT-6 Sol’s best score by 6.4 percentage points at a lower reasoning effort and cost. On GDP.pdf, which measures how accurately models answer professional questions using complex PDF documents, the company reports the model scores higher than Opus 5.5 with fallbacks at less than half the cost per task, and approaches Astra’s performance at roughly one-fifth of the cost per task.

On AutomationBench, which tests whether agents correctly complete multi-step business workflows, OpenAI reports GPT-6.1 Sol scores 2.2 percentage points above Opus 5.5 at medium reasoning effort and at roughly a third of the cost, and 4.8 percentage points above GPT-6 Sol at the same setting. On OSWorld 2.0’s offline set of computer-use workflows, the company reports the model outperforms GPT-6 Sol by seven percentage points at maximum reasoning effort and comes within 2.1 percentage points of Astra’s score at roughly one-seventh of the cost per task.

On Terminal-Bench Science 0.1, which covers scientific workflows including data analysis, simulation, and theorem proving, OpenAI reports GPT-6.1 Sol more than doubles GPT-6 Sol’s score at maximum reasoning effort while averaging $5.47 per task, compared with $23.21 for Opus 5.5 and $23.80 for Astra. The company notes GPT-6 Astra still achieved the highest score among the models tested, at 68.1%. OpenAI states that evaluations of its own models were performed in its research environment or through its API, and that competitor-model evaluations were taken from publicly available reports.

On factuality, OpenAI reports the share of responses containing a factual error fell from 11.4% with GPT-6 Sol to 7.7% with GPT-6.1 Sol at low reasoning effort, a reduction of approximately 32%, and that the new model’s error rate stayed within 1.9 percentage points of Astra’s across the tested settings. The company notes the evaluation uses de-identified conversations in which users had flagged an earlier model’s error, and that these deliberately difficult prompts are not representative of typical usage.

OpenAI also reports improved alignment evaluation results over GPT-6 Sol, with lower failure rates on transparency about broken tools, respecting explicit restrictions, and avoiding unauthorized outcomes during agentic tasks. In an evaluation of whether agents disclose a broken search tool rather than give a best guess, the company reports GPT-6.1 Sol failed to disclose the problem in 2.1% of cases, compared with 4.9% for GPT-6 Sol and 1.5% for Astra, and says it observed no attempts to bypass an automated safety reviewer. Full details appear in a GPT-6.1 Sol system card addendum.

New Developer Tools in Codex and the API

Codex can now run on a computer, remotely from a phone, or in the cloud from any device, with reusable development environments that give teams a shared setup with approved settings and permissions; cloud access is available on Plus, Pro, Business, Healthcare, Education, and Enterprise plans. A refreshed Codex CLI adds voice steering and an /agents view for delegating and tracking multiple tasks, and a new Code Review experience in the ChatGPT desktop app can take an automatic first pass on changes in the cloud, with feedback shared on GitHub pull requests or GitLab merge requests. Both are available on all plans.

Codex Security Cloud scans GitHub repositories on demand or on a schedule, investigates findings, removes duplicates, and prepares fixes in the cloud, and includes access to models offered through Daybreak Blue without a separate Daybreak application. It is available to Pro, Business, Enterprise, and Edu users on desktop and web.

The Decisions API applies Luna’s intelligence to user-defined questions with finite pre-defined answers, letting developers classify content, route requests, or choose an agent’s next action; it entered limited preview on September 29, 2026 with a broad release planned in the coming days. The Agents API now supports computer use and brings Codex’s multi-agent capabilities, tool search, tool calling, and context compaction into applications; it is available through the API and in Codex and ChatGPT Work on Pro 500 and Enterprise plans. OpenAI also worked with Amazon on Bedrock Managed Agents, which adapts the Agents API’s core capabilities to run natively in AWS and integrate with AWS resources.

Ultrafast, a premium speed tier, offers up to eight times faster token generation, at 300 tokens per second, in Codex and up to six times faster generation in the API, according to OpenAI. GPT-6 Astra Ultrafast is available now in the API and in ChatGPT Work and Codex on Pro 500 and Enterprise plans, with GPT-6.1 Sol Ultrafast coming soon.

ChatGPT Collaboration, Plugins, and Plans

ChatGPT Space is a shared home where teammates, ChatGPT, and a user’s dot can build on shared knowledge, available on Pro, Business, and Enterprise plans on desktop and web, with mobile limited to finding, reading, and sharing pages until mobile creation and editing arrive. Pages, a new document type built for human and agent collaboration, are available on the same plans, and collaborative slides with simultaneous multi-editor support and export to PowerPoint or Google Slides are due in the coming weeks.

Business and Enterprise users can now create teams in ChatGPT and delegate recurring team tasks that run on a schedule or respond to events such as a new email or Slack message, and they can mention @ChatGPT in Slack and Microsoft Teams channels, threads, or direct messages without each teammate needing an individual license. The Meetings plugin, in beta in the ChatGPT desktop app on macOS for Pro and Business users with Enterprise access coming, takes notes and saves personalized summaries with action items in ChatGPT Space, deleting the audio once notes are ready. Shareable profiles that showcase reusable Sites and plugins are available across Free, Go, Plus, Pro, Business, and Enterprise plans, with skills sharing limited to the workspace.

For developers, plugin extensions add sidebar homes, interactive panels, and file viewers inside ChatGPT, and OpenAI added a Plugin Creator tool, a redesigned submission flow, improved plugin ranking and discovery, plugin hosting inside Sites on Business, Enterprise, Healthcare, and Edu plans, and support for the proposed MCP Events specification so plugins can start automations when something happens in a connected app.

Sign in with ChatGPT lets Plus and Pro users apply their plan allowance across 16 partners, including Cognition’s Devin, Notion, Vercel, T3, OpenClaw, and Dactyl, with per-tool usage controls, while identity sign-in is available globally. A new Pro 500 plan offers what OpenAI describes as its highest usage allowance at 25 times the ChatGPT Plus allowance, includes Ultrafast access, and is available now. The OpenAI Marketplace lets eligible enterprise customers apply part of their existing OpenAI commitment toward approved partner software, with the first 32 partners including Adobe, Figma, Sierra, Decagon, Hubspot, Salesforce, ServiceNow, Harvey, Legora, Palo Alto Networks, CrowdStrike, and Baseten, and enterprise customers can express interest.

Under the name OpenAI Private Intelligence, the company offers Zero Data Retention with Private Safety Processing, which it says enables automated safety reviews without giving OpenAI personnel access to the underlying content, and plans a Private Inference preview combining confidential computing with strict, verifiable controls for this fall. OpenAI also introduced Dots, always-on agents available on Pro and Business Premium plans in eligible markets, with Enterprise, Edu, and Healthcare access as an admin-enabled beta that is off by default.