On this page
On November 18, 2026 at 01:45:00 Asia/Shanghai (OpenAI News RSS pubDate; social rollout confirmed at 01:54:03 Asia/Shanghai, with API changelog dated November 17), OpenAI officially released GPT-6.2 Astra (gpt-6.2-astra) and GPT-6.2 Sol (gpt-6.2-sol)—superseding earlier pre-listing Codex usage sightings from November 1.

Image source: OpenAI Blog (https://openai.com/index/introducing-gpt-6-2-astra-sol/).
Product Availability & Ultrafast Expansion
**ChatGPT Work, Codex & Pro 200 Ultrafast Rollout: **GPT-6.2 Astra targets complex, high-stakes projects while GPT-6.2 Sol optimizes speed and cost-efficiency for everyday workflows, with OpenAI noting both deliver stronger results using fewer output tokens. Both models are available across ChatGPT Work and Codex for Plus, Pro, Enterprise, and Edu users (workspace admin enablement required for eligible Enterprise/Edu plans). Pro 500 subscribers can use Ultrafast on either model, and OpenAI is bringing Ultrafast to Pro 200 subscribers for the first time, starting with GPT-6.2 Sol.
API Specifications & Official Pricing Tiers
**Context Window (1.05M) & Responses API Requirement: **Both models support text and image inputs with text output, a 1,050,000-token context window, 922,000 max input, 128,000 max output, and reasoning tokens (requests with reasoning effort set to none are rejected). Knowledge cutoffs are May 18, 2026 for Astra and June 1, 2026 for Sol. While both Chat Completions and Responses APIs are supported, tool calling requires the Responses API. The public model-card table lists Standard rate limits for Build (5,000 RPM / 1M TPM), Launch (10,000 RPM / 4M TPM), and Grow (15,000 RPM / 40M TPM); higher tiers are not listed there and are not inferred here.
**Standard, Long-Context, Ultrafast & Background Windows: **For short context (≤272K input tokens), Standard pricing is $8.00 input / $40.00 output / $0.40 cached input for gpt-6.2-astra (20% lower input/output and 60% lower cached input than GPT-6 Astra) and $2.00 input / $10.00 output / $0.10 cached input for gpt-6.2-sol. Above 272K input tokens, input and cache rates double and output is 1.5x short-context rates ($16.00/$60.00/$0.80 for Astra; $4.00/$15.00/$0.20 for Sol). Batch runs at 50% of Standard, Fast at 2x Standard, and Ultrafast is priced at 3× Standard through February 28, 2027 PT ($24.00/$120.00/$1.20 short-context for Astra; $6.00/$30.00/$0.30 for Sol) and at 6× Standard list price from March 1, 2027 PT. Discounted background completion windows in public beta offer 30% off Standard for 2-hour windows and 40% off for 8-hour windows. Separately, the official pricing table adds 10% for regional data residency and FedRAMP endpoints on models released on or after March 5, 2026.

Image source: compiled by DigIntoAI from OpenAI official API pricing tables and model cards.
Vendor Evaluations & Astra for Law
**Self-Reported Benchmarks & gpt-6.2-astra-law: **In OpenAI's evaluations, GPT-6.2 Astra leads GPT-6 Astra by 5.2 points on Agents' Last Exam at under one-third the cost per task and leads Claude Fable 5.2 by 5.5 points (noting Fable 5.2's leaderboard score excludes nine tasks that used a fallback model), gains 9 percentage points over GPT-6 Astra on Terminal-Bench-Science 0.1, and beats GPT-6 Astra by 6.2 percentage points on internal GDPval v2 with 22% fewer output tokens. GPT-6.2 Sol lands within 0.5 points of Astra on FrontierCode 2.0 at roughly one-fifth the cost and outperforms GPT-6 Astra on LifeVal at about one-seventh the cost. On internal GDPval v2, Sol leads GPT-6.1 Sol by 5.3 points. Cybersecurity evaluations were conducted using internal configurations with safeguards and classifiers disabled (detailed in the system card). Alongside the main launch, OpenAI introduced gpt-6.2-astra-law (dated November 17, 2026) for Enterprise Trusted Access and API applicants, priced at $8.00/$40.00/$0.40 short-context with promotional pricing running at least through December 31, 2026.
