OpenAI released GPT-6.1 Sol in the API, ChatGPT Work and Codex. Standard prices are $2 per million input tokens, $0.10 cached input and $10 output. OpenAI reports near-GPT-6 Astra capability on coding, computer use and professional work at about one-fifth of Astra token prices. The Responses API also exposes beta multi-agent delegation.
OpenAI releases GPT-6.1 Sol near Astra capability at lower cost
2026-09-29Verified announcementTRIAL
OpenAI offers GPT-6.1 Sol in the API, ChatGPT Work and Codex at a lower standard token price than Astra. The API now makes Ultrafast processing generally available through service_tier: ultrafast, with separate limits, global processing and US or EU data residency.
Official · Referenceopenai.com · introducing gpt 6 1 solOfficial · Referencedeploymentsafety.openai.com · awareness of cot monitoring reduces monitorability
Availability: OpenAI API, ChatGPT Work and Codex; not Chat at launch
What this means for you
For latency-sensitive agents, Ultrafast is now a production-available routing choice rather than a limited preview. Measure end-to-end task latency and completed-task cost under your residency constraints before changing the default tier.
Evidence & related material
The OpenAI API changelog confirms GPT-6.1 Sol Ultrafast is available to all API users.
OpenAI staff report Ultrafast reducing some security-incident investigations from one or two hours to 10–15 minutes.
Subsequent changes
GPT-6.1 Sol Ultrafast becomes available to all API users
OpenAI reports an internal incident-response use case for Ultrafast
Technical details & scores
| Model | gpt-6.1-sol |
|---|---|
| Availability | OpenAI API, ChatGPT Work and Codex; not Chat at launch |
| Pricing Per Million Tokens | input 2 cached_input 0.1 cache_write 2.5 output 10 |
| Context Pricing Threshold | standard prices shown for prompts up to 272K input tokens |
| Vendor Results | matches Astra on DeepSWE v1.1 at about one-fifth cost; within 2.1 points on OSWorld 2.0 at about one-seventh task cost |
| Multi Agent | beta delegation in Responses API |
| Safety | Critical cybersecurity and High biological/chemical capability under OpenAI Preparedness Framework |
| Ultrafast 2026 10 08 | Generally available to all API users with service_tier: ultrafast; separate rate limits; global processing and US/EU data residency options. |
| Internal Incident Response 2026 10 26 | OpenAI staff report investigations falling from 1–2 hours to 10–15 minutes; vendor internal case study without controlled comparison |
Technical impact 5
Engineering value 5
Adoption signal 4
Maturity 4
Verification cost 3
Risk 4