AI Launch Profile

GPT-6 Sol

OpenAI added GPT-6 Sol to the API as gpt-6-sol and began rolling it out in ChatGPT Work and Codex to eligible paid plans on September 22, 2026.

Read official source Tool profile

At a glance

Launch Snapshot

Company: OpenAI
Launch date: September 22, 2026
Launch type: Model Release
Category: AI Models
Audience: Developers
Pricing: OpenAI lists standard direct API rates of US$2 per million input tokens and US$10 per million output tokens for requests with at most 272K input tokens, checked September 22, 2026. Longer context, caching and processing tiers have different rates; consult the official model pricing.
Free plan: No
API: Yes
Open weights/source: No

Launch Context

Use these links to move from this record into the broader Launch Intelligence database.

Launch Intelligence hub Today's launches This week's launches More: AI Models More: Model Release

Verification & Sources

Evidence state: Recheck due
Source links: 3
Freshness: Needs recheck: checked September 22, 2026
Last updated: October 9, 2026

What this evidence state means

Definition: The claim was previously checked, but its review window expired or a material change may have invalidated it.
Required provenance: The prior evidence and check date are retained, together with the expiry or change signal that triggered recheck.
Owner: Kingy freshness queue owner and assigned editorial reviewer
Freshness rule: This is already outside its freshness rule. It must not be presented as current until reviewed against current evidence.
Disputes and corrections: Use “Suggest a correction” on the record. Kingy editorial reviews the cited evidence, records material corrections, and changes or removes the state when it is not supported.

Key source checks

Official site Pricing Source 3

Suggest a correction

Kingy AI Take

GPT-6 Sol merits a workload-specific trial when Astra costs more than the task warrants. Kingy's September 22 comparison finds lower cost per task on OpenAI's reported tests, while noting that Sol's top scores trail GPT-5.6 Sol on DeepSWE and OSWorld 2.0. Those scores are vendor-reported, not Kingy hands-on results.

Who it is for

Developers

What feels promising

The model gives developers another price and capability point for sustained coding and agent tasks. Its value depends on the workload and reasoning effort; vendor evaluations do not establish a universal winner.

What feels unproven

OpenAI's launch-day benchmark figures are vendor-reported and may differ from production ChatGPT output. GPT-6 Sol is not the highest-scoring model on every reported task. Availability may vary during the gradual rollout.

Source list

Sources

Official site openai.com Source developers.openai.com Source kingy.ai