GPT-6 Astra (Organization): news timeline & CVEs

How invideo improves color grading 3x with GPT‑6 Astra

Agentic video editor invideo reports GPT-6 Astra roughly tripled color-grading success rates and improved frame-level edit planning.

Invideo, an agentic video editing platform, says GPT-6 Astra plans complex edits at frame-level accuracy using fewer reasoning steps and output tokens than previously tested models. CEO Sanket Shah reported color-grading and color-correction success rates improved about threefold with Astra. A few editors also used the model to build roughly 50 custom, fully editable effects in a single day.

OpenAI NewsSep 23, 12:00 UTC · 4d agoAI industry 2 sources

Harvey turns legal context into stronger drafts with GPT-6 Astra

Legal AI firm Harvey uses GPT-6 Astra to draft more complete legal documents with better formatting and lawyer preference controls.

Harvey, an AI platform for law firms and in-house teams, reports substantial improvements in document formatting and context awareness using GPT-6 Astra for litigation and M&A drafting. A memory panel lets lawyers encode preferences such as numbered lists, prioritizing EDGAR as a source, and color-coding issues by priority alongside source material and draft memoranda. Cofounder and President Gabe Pereyra says feeding more context to the model produces better structured outputs.

OpenAI NewsSep 23, 12:00 UTC · 4d agoAI industry

OpenAI Releases GPT-6 Sol and Luna: 50% Cheaper API Pricing and Benchmarks

OpenAI launched GPT-6 Sol and Luna API models at roughly half of GPT-5.6 pricing, with Sol beating Claude Opus 5 on several benchmarks.

Sol ($2/$10 per 1M input/output tokens) and Luna ($0.10/$0.50) are API-only models positioned below GPT-6 Astra, targeting complex coding and fast high-volume work respectively. Sol scores 33.2% on AutomationBench 1.0.6 versus Claude Opus 5's 26.9% at 11x the cost, plus 56.4% on Agents' Last Exam, 68.8% on DeepSWE v1.1, and 60.5% on OSWorld 2.0. GPT-6 also ships improved prompt caching with up to 90% discounts on cached input reads, a caching dashboard, and new cache-control developer features.

MarkTechPostupdated Sep 25, 05:37 UTC · 2d agofirst Sep 23, 05:18 UTC · 4d agoModel release 12 sources

New Anthropic, OpenAI models make same promise: A little more for a lot less money

Anthropic's Opus 5.5 and OpenAI's GPT-6 Sol and Luna offer modest capability gains at roughly half the cost of predecessors.

Anthropic claims Opus 5.5 delivers about 40 percent savings versus Opus 5 on typical workloads through cheaper tokens and lower token usage, and says it is notably capable in high-risk areas like cybersecurity and biology, with flagged requests routed to older models. OpenAI released GPT-6 Sol and Luna, trained with methods similar to GPT-6 Astra, scoring a few percentage points better on some benchmarks at half the cost. Sol is positioned as the efficient daily driver and Luna as the fast, cheap tier in the GPT-6 family.

Ars Technica · AIupdated Sep 25, 05:37 UTC · 2d agofirst Sep 22, 21:25 UTC · 4d agoAI industry 12 sources

OpenAI GPT–6 Astra breaks Enigma message that has resisted solution since 2005

OpenAI's GPT-6 Astra autonomously broke the unsolved 1941 Enigma message MVUEH, writing its own simulator, Bombe code and archive research.

OpenAI's GPT-6 Astra independently solved Enigma message MVUEH (Nr. 172, 10 July 1941, 82 letters), unbroken since 2005, after writing its own Python and C++ Enigma simulator and Bombe and using the repeated crib ROSENOW. The recovered key used wheel order 253, differing from the daily key 512, and validation by Frode Weierud of Crypto Cellar Research confirmed the plaintext, nearly identical to message SIPVX (Nr. 173). The model also traced correct Bundesarchiv file references (RS 3-3/20a, RS 3-3/63b), accomplishing in two days work that took humans weeks or months.

Hacker News · AISep 22, 13:52 UTC · 4d agoAI research HN 71↑ · 82 comments1· 1 read

GPT-6 Astra

Timeline5 articles

Appears with

How invideo improves color grading 3x with GPT‑6 Astra

Harvey turns legal context into stronger drafts with GPT-6 Astra

OpenAI Releases GPT-6 Sol and Luna: 50% Cheaper API Pricing and Benchmarks

New Anthropic, OpenAI models make same promise: A little more for a lot less money

OpenAI GPT–6 Astra breaks Enigma message that has resisted solution since 2005