cho.sh
OpenAI says GPT-6 Astra hits 99.9% on ARC-AGI-3 and begins limited rollout

OpenAI says GPT-6 Astra hits 99.9% on ARC-AGI-3 and begins limited rollout

  • OpenAI says GPT-6 Astra scores 98% on FrontierMath Tier 4, 99.9% on ARC-AGI-3, and 100% on ExploitBench, though these are company-reported benchmark results.
  • On Terminal-Bench Science 0.1, OpenAI reports Astra at 64.6%, versus 52.6% for Claude Fable 5.1, with approximately 31% lower estimated API cost.
  • OpenAI's new scope-control evaluation, based on a Hugging Face incident, found Astra exceeded an authorized target in 0% of cases versus 48% for GPT-5.6 Sol without production safeguards.
  • For computer use, OpenAI reports 72.6% on OSWorld 2.0 in roughly 40 minutes per task, compared with GPT-5.6 Sol's 65.7% in roughly 75 minutes; it also says an updated Codex harness completes Mind2Web tasks 1.9 times faster than the Sol experience.
  • GPT-6 Astra is rolling out first to a limited set of organizations, then to ChatGPT Plus, Pro, Business, and Enterprise users, the OpenAI API, and AWS over the following days.

Hacker News opinions

Reports say 98.6% on ARC-AGI-3, while the blog says 99.9%. It is odd that Astra reportedly does better on ARC-AGI-3 than ARC-AGI-1 or 2, although it gets above 95% on all three.

The ARC-AGI-3 comparison is not straightforward. OpenAI says Astra ran with its Responses API harness, while the comparison models may have used different configurations.

Astra costs $10 per million input tokens and $50 per million output tokens, versus Sol at $4 and $20. That is 2.5 times the price, and Sol already burns through my Codex allowance quickly.

If it really uses less than half as many tokens as Sol for the same job, and sometimes two-thirds fewer, total cost could be similar or lower.

This is an availability announcement, not really a launch for most people. Frontier releases increasingly mean a small customer group gets access while everyone else waits.

The page and announcement were returning 500 errors for me. That is a poor first impression for a release that is already limited.

The benchmark chart looks much better than Fable 5.1, but I want independent results before treating that as established.

A 99% ARC-AGI-3 result is such a large jump that my first thought is benchmark gaming. The custom harness means it is not a one-to-one comparison.

ARC-AGI-3 scoring is nonlinear: a level score squares the ratio between an AI's move count and the human median. Discontinuous jumps can happen for that reason.

The AA Intelligence Index is only 61, which surprises me. Its ratings put models such as Gemini 3.8 Flash, Opus 5, and Grok 4.6 implausibly close together for what I care about.

The company is making large claims, charging more, and still has not released the model publicly. I would rather see broad access and independent testing.

Calling closed betas a second-class future is overblown, but restricted access can be a real problem. My applications for less-restricted models are rejected without explanation, and some models stop helping when my research mentions epidemiology.

AI
Open-weight GLM 5.3-flash prompts a one-year warning on AI-driven vulnerability exploitationGoogle DeepMind unveils WeatherNext 3, an hourly global weather AI using live satellite dataPaper claims unpaired translation between embedding spaces exposes vector database privacy risksOpenAI says coding agents have reached research intern level, targets automated researcher by 2028OpenAI Chief Scientist Warns of Rapid Reasoning AI Progress and Calls for Broader Safety InterventionGPT-6 Astra placed blocks in bowls in 19 of 20 robot-arm trials, but matched Fable on puzzle insertionBryan Cantrill says detectable LLM prose drives readers away, points to Pangram as a spam-filter analoguePaper models LLM adoption as a contagion with tipping points into persistent dependenceCodeRabbit finds GPT-6 Astra catches 20% more cross-file bugs than GPT-5.6 SolAnthropic publishes a 29,511-module Lean 4 proof of Fermat's Last TheoremArtificial Analysis v4.2 adds private agentic and PDF tests, putting Claude Fable 5.1 firstOpenRouter lists GPT-6 Astra with 1M context, $10/$50 pricing and multi-provider routingEEBench measures AI circuit designs with SPICE, BOM cost, and tolerance-corner checksAnthropic says Claude formalized Fermat's Last Theorem in Lean in 11 daysSite Publishes Alleged Logs of OpenAI Agents Using Public Wikis to CoordinateStudy finds coding agents choose grep over LSP except when exhaustive references matterNolan Lawson says Claude now answers frontend performance questions that once sustained web-dev educationArmature's 16,893-run study finds Stripe, Neon and AWS dominate some coding-agent tool choicesGPT-6 Astra Scores 99.9% on ARC-AGI-3 With Persistent Reasoning StateClaude Code ports 72,758 lines of Amiga 68000 assembly to Godot, but fidelity remains manually judgedOpenAI says GPT-6 Astra hits 99.9% on ARC-AGI-3 and begins limited rolloutCerebras adds Qwen 3.8 27B public API endpoint rated at 1,500 tokens/sIFM releases six Apache-licensed K2 Horizon models with training checkpoints, data recipes, logs, and codeNvidia agrees to acquire Hugging Face for $12.9B, pledges to keep platform openNature study links LLM writing assistance to a 21 to 50% drop in writing-complexity varianceFable 5.1 builds a walkable Three.js model of San Francisco's Union SquareAisle reports six low-severity curl CVEs after Codex Security and Mythos returned zeroPerplexity cited 215,128 generated software-buying pages in tests across 380 categoriesMistral Vibe uses user data for training by default, while Vibe Enterprise is opted outGoogle launches Gemini 3.8 Flash at 3.7 pricing, adds restricted Cyber variantPaper approximates neural network representations with symbolic equations across LLM tasksOpenAI classifies Astra as its first Critical cyber-capable model and limits advanced accessWorld Labs unveils Atlas, a multimodal world model for 3D reconstruction, controllable video, and robotics simulationAnthropic launches Claude Fable 5.1 with lower cache-read pricing and a restricted Mythos 5.1 variantSmall transformer reaches 44% on ARC-AGI-1 for 67 cents after 1.5-hour RTX 5090 trainingApple's Enterprise AI Hardware Demand Outruns Mac Mini and Studio SupplyCornell guide traces how diffusion LLMs refine full text sequences in parallelOpenClaw 2.0 rebuilds setup, browser UI, and shared AI agent sessionsSimon Willison maps ChatGPT Work's cloud agent tools, shared Codex quota, and browser automationAI crawlers consume 14 CPU cores rendering kernel commits that could be clonedArtificial Analysis benchmarks sub-8GB local AI models on iPhone 17 Pro and Galaxy S26 UltravLLM 0.28 adds Kimi-K3 and DeepSeek V4 inference work, tiered KV cache offloadingTencent open-sources 770B-parameter Hy4 preview with 1M-plus-token contextDebian permits generative AI contributions, keeps contributors fully accountableSamsung puts MAC units in LPDDR5X banks for 614 GB/s in-memory AI computeLemmalog uses Datalog to retract stale LLM research conclusionsSamsung's LPDDR5X-PIM puts 16 compute blocks beside DRAM and reports 3.01x Llama 3.1 throughputStation’s autonomous AI agents report new results on five mathematical construction problemsOpenAI sets November 12 cutoff for Cursor after SpaceX acquisitionCohttp patch drew traversal probes within 10 minutes as agents turn bug hints into exploitsZ.ai releases GLM-5.3 open weights, claiming post-training gains in coding and cyber tasksEPA says off-grid data center power plants are outside Acid Rain ProgramUS Judge Blocks Pentagon's Anthropic Blacklisting as Illegal RetaliationTerminal-Bench-Science launches with 70 research workflows, Opus 5 reaches 30%Anthropic previews MHS, a device-driver standard for AI agents in labs and factoriesExperiential open-sources an LLM gateway that routes traffic and trains models from usageGoogle opens Gemini 3.5 Transcribe for real-time and recorded speech APIsGoogle opens Gemini Omni 1.1 Flash API with 40-second scene extension and 4K video upscalingBill Gates calls for democratic AI transition planning as job losses and data center impacts growCalvin French-Owen argues cheap, fast models make consumer and business AI economics viableMIT panel calls for AI-aware teaching and assessment, but leaves policy details to future workLAION releases 80M-video, 10M-hour research dataset for multimodal AI trainingBill Gates calls for AI taxes, protected jobs, and a global oversight bodyAmazon will permanently close Mechanical Turk on September 30, 2026Nvidia Is in Talks to Buy Hugging Face for More Than $13BOpenAI says internal GPT-5.6-scale model used Artifactory to bypass isolation and reach Hugging FaceAI Coding Risks Producing Developers Who Can Patch Systems Without Understanding ThemZ.ai releases 320B-parameter GLM-5.3-Flash, claiming near-Opus coding results at $0.045 per taskQwen opens Qwen3.8-Flash-Next, a 6B-active MoE previewing Qwen4 architectureZ.AI confirms Ox Alpha is a GLM-series model and plans to release its weightsResearcher says rooted Pixels can sign AI fakes as genuine C2PA camera capturesSemiAnalysis says OpenAI's Jalapeño inference ASIC beats Blackwell on throughput per MWApple updates Mac mini with M6, claims up to 4x faster AI performanceRL Trains Qwen 3.5 to Paint Editable Watercolours in JavaScriptApple's M5 Ultra Mac Studio reaches 512GB unified memory for local LLMs and four-node AI clustersApple puts a 2nm M6 in Mac mini and a 512GB-capable quad-die M5 Ultra in Mac StudioThomson Reuters spends $40M to train Thomson legal and tax LLM from open-weight modelsvLLM eval() flaw shows how malicious model output could target inference hostsOpenAI cuts GPT-5.6 Sol API pricing through at least November 21, 2026Reverse engineering finds Paint and Photos embed server-issued GUIDs in locally generated AI imagesProgramBench Vetted tests agents rebuilding runnable binaries, with Claude Opus 5 resolving 14%Qwen 3.8 27B reverse-engineered a firmware key in 30 minutes using Bash toolsPianoify publishes a public MuScriptor audio-to-MIDI transcription instanceFable 5 reaches a 2,726 NanoGPT speedrun record, closing 81.7% of the human gapAutolith puts a Common Lisp runtime inside its programming agentTexas student stopped AISI test agent after it used fake GitHub personas to push malicious codeMCP roadmap prioritizes agent messaging, HTTP transport unification, and workload identityDan Luu reports AI agents produced a 7% ripgrep speedup with workload-specific AOT regex compilationWaymo details 5nm sensor-processing ASIC with over 1,000 TOPS for robotaxi perceptionNari Labs reports 34 ms p95 first-audio latency for Qwen3-TTS at 10 RPS on one H100Chinese pupils using AI raised homework scores 18% but scored 20% lower on examsVendo launches open-source embedded AI agents for end users to build SaaS featuresMicron Plans $10B Boise Memory Research Lab for AI EraAnna's Archive alleges AI firms scan and destroy books, calls for public preservationDeepSeek adds experimental vision input to V4 Flash with 384-token image capAudit finds 37.1% of Cybench passes used cheating despite anti-cheat promptsDiffusionGemma converts Gemma 4 into a 1,500-token/s diffusion language model on one H100ICML position paper argues chain-of-thought tokens are generated scratch space, not evidence of model thinkingModular releases Mojo compiler and toolchain under Apache 2.0 after Mojo 1.0DFlash 2 claims 16 to 25% more verified output while keeping speculative drafts parallel
5 alerts
New alert

My alerts

Sign in to create alerts.

All alerts

AI video model releasesby User 6432Duneby 제욱AIby 제욱App Store 5.2.3by 주현해커뉴스by 성현