OpenAI released GPT-6.2 Astra and GPT-6.2 Sol today, November 17, 2026, after pulling an earlier version to strengthen safety safeguards. GPT-6.2 Astra is now available to ChatGPT Plus, Pro, Enterprise, and Edu subscribers through ChatGPT Work and Codex, according to OpenAI.
What OpenAI announced: GPT-6.2 Astra and Sol
GPT-6.2 Astra cuts API prices by roughly 20% versus GPT-6 Astra: input drops from $10 to $8 per million tokens, output from $50 to $40 per million tokens, and cached input from $1 to $0.40 per million tokens (a 60% reduction), according to OpenAI's pricing page and data tracked by Artificial Analysis. GPT-6.2 Sol keeps GPT-6.1 Sol's rates: $2/M input and $10/M output. Important caveat: for requests exceeding 272,000 input tokens, OpenAI doubles input and cache rates and applies a 1.5× multiplier to output rates for the entire request. On benchmarks, OpenAI reports that Astra scores 5.2 points above GPT-6 Astra on Agents' Last Exam at less than one-third the per-task cost, and that GPT-6.2 Astra ranks second on the Artificial Analysis Intelligence Index (AA-3D score: 1,318 vs. 1,271 for Claude Fable 5.2). These figures come from OpenAI; fully independent benchmarks are not yet available.
"A 20% price cut on OpenAI's flagship model, paired with reported gains on multistep tasks, opens concrete conversations for small and mid-sized businesses that need to automate complex workflows without blowing their API budget."
Davarion Group & LabsWhat GPT-6.2 Astra and Sol mean for SMBs
- 01Lower entry cost: at $8/M input tokens (down from $10), document processing, contract review, and email workflows become less expensive — actual savings depend on volume and system design.
- 02GPT-6.2 Sol as a budget-friendly tier: at $2/M input and $10/M output, Sol offers a mid-range option well suited for repetitive tasks like classification, summarization, or first-level customer responses.
- 03Watch the long-context pricing spike: requests above 272,000 input tokens face doubled rates; systems handling very long contexts (legal files, lengthy contracts) need to be carefully architected or budgeted.
- 04Test before migrating: current benchmark data comes primarily from OpenAI; run your own evaluations on real use cases before rerouting production traffic to GPT-6.2 Astra.
Teams already running GPT-6 Astra in production may find the price reduction creates headroom to increase request volume or reinvest in other automation layers. For teams considering a first deployment, pairing Astra (complex tasks) with Sol (routine tasks) offers a two-tier setup that can be sized to a small business budget.
At Davarion Group & Labs, we help Houston-area businesses assess which model fits their real workflows — from reviewing technical requirements and estimating API cost impact, to designing and implementing automation pipelines. If you want to explore whether GPT-6.2 Astra or Sol belongs in your operation, reach out at davarion.com.