RELEASE · MODELS · #3760
OpenAI introduces GPT-6.2 Astra and Sol
OpenAI announced GPT‑6.2, released as two variants — Astra and Sol — promising improved professional outputs, coding, and computer use with greater efficiency and lower cost across ChatGPT Work, Codex, and the API. The update is positioned to deliver better performance for enterprise and developer workflows.
KEY POINTS
- OpenAI announced GPT‑6.2, released as two variants — Astra and Sol — promising improved professional outputs, coding, and computer use with greater efficiency and lower cost across ChatGPT Work, Codex, and the API.
- The update is positioned to deliver better performance for enterprise and developer workflows.
- A new GPT‑6.2 release that improves capability and cost-efficiency across ChatGPT Work, Codex, and the API can materially affect enterprise adoption, developer tools, and downstream AI products.
DECISION BRIEF
CONFIDENCE
WHAT CHANGED
OpenAI announced GPT‑6.2 delivered as two variants, Astra (positioned for demanding professional work) and Sol (positioned for everyday tasks). OpenAI and press reporting say the release promises improved professional outputs, coding, and computer use with greater efficiency and lower cost across ChatGPT Work, Codex, and the API; the models are available now to Plus, Pro, Enterprise, and Edu users and through the API, with stated price reductions for Astra versus its predecessor.
WHY NOW
The update is presented as immediately available and includes explicit API price changes (Astra priced lower than its predecessor and Sol retaining prior pricing) plus product availability and Ultrafast access adjustments for Pro subscribers — factors that can change near-term cost/performance choices for developers, enterprise users, and paid subscribers.
WHO IS AFFECTED
Directly affected per the sources: Plus, Pro, Enterprise, and Edu ChatGPT users; Pro-500 and Pro-200 subscriber tiers (Ultrafast mode access changes); developers using the OpenAI API; professional workflows in fields named by OpenAI/press (science, law, finance, accounting) that Astra is trained to target.
CONFIRMED
Explicit source-backed facts: - OpenAI announced GPT‑6.2 Astra and GPT‑6.2 Sol (OpenAI press excerpt). (source claim) - OpenAI/press state the models offer better professional outputs, coding, computer use, and greater efficiency at lower cost across ChatGPT Work, Codex, and the API (OpenAI excerpt; The Decoder summary). (source claim) - The Decoder reports Astra is designed for demanding specialized work (science, law, finance) and Sol is geared toward quick, everyday tasks. (The Decoder) - The Decoder reports both models are available to Plus, Pro, Enterprise, and Edu users and accessible via the OpenAI API. (The Decoder) - The Decoder reports Pro-500 subscribers can use Ultrafast mode with both models and Pro-200 subscribers gain Ultrafast access starting with GPT‑6.2 Sol. (The Decoder) - The Decoder gives Astra API pricing as 20% cheaper than its predecessor and lists per‑million token prices: Astra — $8 per million input tokens, $0.40 for cached inputs, $40 per million output tokens; Sol — $2 input, $0.10 cached, $10 output (reported as same as GPT‑6.1 Sol). (The Decoder) - The Decoder reports benchmark claims: Astra ranks second in the Artificial Analysis Intelligence Index (behind Claude Fable 5.2) and is described as dominating the Pareto frontier on cost-performance; on Agents' Last Exam OpenAI says Astra scores 5.2 points above GPT‑6 Astra at less than a third of the cost and 5.5 points above Claude Fable 5.2 at ~5% of the cost; on Harvey LAB‑AA OpenAI says Astra more than doubles Claude Fable 5.2's pass rate at about one-sixth of the cost; OpenAI reports a 6.2 percentage‑point gain on an internal GDPval‑v2 benchmark (excerpt truncated). (The Decoder / OpenAI claims)
UNCERTAIN
Missing or conflicting evidence (explicitly labeled): - Independent verification: many performance and cost comparisons are reported as OpenAI claims or The Decoder summaries; independent third‑party replication details and methodologies for benchmarks (Agents' Last Exam, Harvey LAB‑AA, GDPval‑v2, Artificial Analysis index) are not provided in the supplied excerpts. (missing) - Incomplete data: the GDPval‑v2 result is truncated in the supplied excerpt; full metrics, sample sizes, and task breakdowns are not present. (missing) - Scope and geography: the excerpts do not specify regional availability, rollout timing beyond 'available now', or contract/volume pricing for enterprise customers. (missing) - Real‑world adoption and safety/behavior changes: the supplied excerpts report claimed improvements but do not provide customer adoption metrics, failure modes, latency, or safety/guardrail performance. (missing) - Price applicability: the excerpts state specific API prices and a 20% reduction for Astra versus its predecessor per OpenAI reporting; whether those prices and cached‑input rules apply to all customers, regions, or billing arrangements is not specified. (missing) - Conflicting or partial claims: The Decoder reports Astra ranks second in the Artificial Analysis index while also reporting OpenAI framing Astra as dominating the Pareto frontier on cost‑performance; both can be simultaneously true but the excerpts do not include the underlying data for reconciliation. (incomplete/conflicting context)
WHAT TO WATCH
Concrete observable signals to monitor next (based on supplied sources): - Official OpenAI product and pricing pages or API docs for any updates or clarifications to the per‑token prices and cache pricing for Astra/Sol. - ChatGPT product pages, release notes, or account emails showing availability changes or Ultrafast access changes for Pro-200/Pro-500 tiers. - Independent benchmark publications or replications (Artificial Analysis index updates, Agents' Last Exam results, Harvey LAB‑AA, third‑party tests) that reproduce or challenge the reported cost/performance claims. - Media or vendor reports comparing Astra/Sol to Anthropic’s Claude Fable 5.2 and Google’s Gemini 4 Carbon in real tasks. - Developer/API console listings and SDK docs showing model availability, quotas, and any region/enterprise pricing differences.
WHY IT MATTERS
A new GPT‑6.2 release that improves capability and cost-efficiency across ChatGPT Work, Codex, and the API can materially affect enterprise adoption, developer tools, and downstream AI products.
EVIDENCE MAP
4Editorial claims linked to specific sources, with support, contradiction and context shown separately.
OpenAI announced GPT-6.2, released in two variants called Astra and Sol.
SUPPORTEDChecked 2026-11-17 · 2 supporting
OpenAI says GPT‑6.2 offers better professional outputs, coding, computer use, and greater efficiency at lower cost across ChatGPT Work, Codex, and the API.
SUPPORTEDChecked 2026-11-17 · 2 supporting
OpenAI's excerpt describes better professional outputs, coding, computer use, and efficiency at lower cost across ChatGPT Work, Codex, and the API.
SUPPORTS The DecoderMEDIA · 2026-11-17The Decoder repeats OpenAI's claim that both models outperform predecessors on benchmarks and cost less to run across services.
Astra is positioned for demanding specialized professional work (e.g., science, law, finance), while Sol is positioned for quicker, everyday tasks.
SUPPORTEDChecked 2026-11-17 · 1 supporting
Both GPT‑6.2 Astra and Sol are available to Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex, and developers can access them through the OpenAI API.
SUPPORTEDChecked 2026-11-17 · 2 supporting
OpenAI's excerpt lists ChatGPT Work, Codex, and the API as targets for the GPT‑6.2 update.
SUPPORTS The DecoderMEDIA · 2026-11-17The Decoder reports availability to Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex, and notes developer access via the API.
SOURCES & TIMELINE
2Introducing GPT‑6.2 Astra and Sol, with better professional outputs, coding, computer use, and efficiency at lower cost across ChatGPT Work, Codex, and the API.
OpenAI has released two new models: GPT-6.2 Astra and GPT-6.2 Sol. They are on par with Anthropic’s best models, but much more affordable. Astra is designed for demanding specialized work in science, law, and finance, while Sol is geared toward quick, everyday tasks. Both are now available to Plus, Pro, Enterprise, and Edu users. Astra’s API pricing is 20 percent lower than that of its predecessor. In the Artificia…