Key Takeaways
- GPT-6.1 Sol launched September 29 at $2 (around ₱125) per million input tokens and $10 (around ₱630) per million output tokens — one-fifth of GPT-6 Astra's prices — while nearly matching Astra on agentic coding and computer use.
- The Agents API entered public beta with hosted execution, computer use and multi-agent support.
- Dots, always-on agents that keep working after you log off, rolled out to Pro and Business Premium subscribers in eligible markets; the Philippines is not on a confirmed list.
- An Ultrafast tier generates tokens up to 8x faster in Codex, at 6x the usual price, alongside a new $500-a-month (around ₱31,300) Pro 500 plan.
OpenAI used its DevDay event on September 29 to push the cost of agent-grade AI down and its autonomy up. The headline release, GPT-6.1 Sol, nearly matches the company's flagship GPT-6 Astra on agentic coding, computer use and professional work at one-fifth of Astra's token prices.
GPT-6.1 Sol pricing and performance
The new model costs $2 (around ₱125) per million input tokens and $10 (around ₱630) per million output tokens in the API, with cached input at $0.10 (around ₱6) per million — 95% below standard input pricing. OpenAI also says factuality improved: errors fell from 11.4% to 7.7% at low reasoning effort.
GPT-6.1 Sol is available from September 29 to Plus, Pro, Business, Enterprise and Edu users in ChatGPT Work and Codex, and to developers in the API as gpt-6.1-sol. Unlike some of the day's other launches, OpenAI announced no market restriction for the model itself. It follows the flagship GPT-6 Astra launch earlier this year — and, notably, comes after OpenAI cancelled the planned October release of GPT-6.1 Astra over unmet safety standards, according to Decrypt.
Agents API and Dots, the always-on agents
The Agents API is now in public beta. It gives developers a managed agent runtime with hosted execution, sessions, tools, MCP support, context management and multi-agent capabilities, and it now supports computer use — agents that click, type and navigate software through its interface.
Dots go a step further. Each Dot is an always-on agent powered by GPT-6 Astra, with its own cloud computer and web browser and integrations with more than 4,000 applications. Dots keep working after the user logs off, learn preferences over time, and follow built-in rules about when to act on their own — password changes stay human-controlled. They are available to Pro and Business Premium subscribers in eligible markets, with Enterprise, Edu and Healthcare access as an admin-enabled beta. Free and Plus ($20 a month, around ₱1,250) users are excluded.
Ultrafast, Pro 500 and the rest of the stack
A new Ultrafast tier generates tokens up to eight times faster in Codex — about 300 tokens per second — and up to six times faster in the API, priced at six times normal usage and powered by Cerebras chips. It comes bundled with the new Pro 500 plan at $500 a month (around ₱31,300) and with Enterprise plans.
The day's other announcements included ChatGPT Space, a shared workspace where teams and agents co-edit documents called Pages; a Decisions API that routes simple classification tasks to Luna, OpenAI's cheapest model; Codex Security Cloud, which scans GitHub repositories for vulnerabilities; and an OpenAI Marketplace that lets enterprise customers spend existing commitments on partner software from 32 vendors, including Adobe, Figma and Salesforce.
What Filipino developers and BPOs get
For developers and IT-BPM firms in the Philippines, the practical news is the price: agent-grade coding and computer-use capability at ₱125-class input pricing makes automated workflows meaningfully cheaper to run at scale, and the API carries no announced market restriction. That matters in a market where AI already handles 46% of Philippine customer-service conversations. The caveat is Dots: OpenAI limits them to "eligible markets" and has not confirmed the Philippines, so the always-on agents may not be switchable-on here at launch. The safety context is also worth holding onto — the same week's cancelled GPT-6.1 Astra release and earlier incidents of OpenAI agents probing government sites are a reminder that autonomy is being priced down faster than it is being fully tamed.
