GPT-6 Astra

Leaderboards

Where GPT-6 Astra sits among the top 20 in each category.

By task

Scored from votes on each kind of prompt alone, on the same 1,000-point scale.

Test results

Every test GPT-6 Astra has a published result on, out of 100%, with its rank among the models that took it.

• GPQA

  #1 / 256

  Graduate-level science questions written to be hard to look up.

  96.0%

  Self-reported

• BrowseComp

  #1 / 68

  Finding hard-to-locate facts by browsing the web.

  91.5%

  Self-reported · In the agents score

• DeepSWE 1.1

  #12 / 50

  DeepSWE 1.1 evaluates software engineering agents using the mini-swe-agent harness.

  74.1%

  Self-reported · In the coding score

• Terminal-Bench 4.0

  #4 / 28

  Terminal-Bench 4.0 evaluates AI agents on real-world work in terminal and command-line containerized environments across 66 tasks, with emphasis on science-adjacent and frontier engineering problems. Relative to earlier Terminal-Bench releases, 4.0 increases timeouts and adaptively raises RAM/CPU on selected tasks to reduce harness and resource confounds.

  57.7%

  Self-reported · In the coding score· In the agents score

• ScreenSpot Pro

  #2 / 27

  ScreenSpot-Pro is a novel GUI grounding benchmark designed to rigorously evaluate the grounding capabilities of multimodal large language models (MLLMs) in professional high-resolution computing environments. The benchmark comprises 1,581 instructions across 23 applications spanning 5 industries and 3 operating systems, featuring authentic high-resolution images from professional domains with expert annotations. Unlike previous benchmarks that focus on cropped screenshots in consumer applications, ScreenSpot-Pro addresses the complexity and diversity of real-world professional software scenarios, revealing significant performance gaps in current MLLM GUI perception capabilities.

  92.7%

  Self-reported

• Agents' Last Exam

  #1 / 24

  Agents' Last Exam is a challenging benchmark for AI agents on hard, long-horizon tasks that test sustained reasoning, planning, and tool use, reported with and without tool access.

  59.3%

  Self-reported · In the agents score

• AutomationBench

  #9 / 24

  AutomationBench is a tool-use benchmark that evaluates AI agents on automating real-world workflows, testing their ability to orchestrate tools and complete multi-step automation tasks.

  41.4%

  Self-reported · In the agents score

• FrontierCode 1.1

  #4 / 23

  FrontierCode 1.1 evaluates whether coding-agent changes are mergeable, using unit tests, maintainer-defined rubrics, and verifiers. Runs flagged for unfair internet use receive a zero score.

  53.3%

  Self-reported · In the coding score

• ARC-AGI v2

  #1 / 19

  ARC-AGI-2 is an upgraded benchmark for measuring abstract reasoning and problem-solving abilities in AI systems through visual grid transformation tasks. It evaluates fluid intelligence via input-output grid pairs (1x1 to 30x30) using colored cells (0-9), requiring models to identify underlying transformation rules from demonstration examples and apply them to test cases. Designed to be easy for humans but challenging for AI, focusing on core cognitive abilities like spatial reasoning, pattern recognition, and compositional generalization.

  95.0%

  Self-reported · In the text score· In the vision score

• HealthBench Professional

  #6 / 19

  HealthBench Professional evaluates model capability and safety for clinician use cases using real clinician-style chats and physician-authored grading rubrics.

  63.4%

  Self-reported

• OSWorld 2.0

  #3 / 17

  OSWorld 2.0 is a benchmark of 108 long-horizon, real-world computer-use workflows spanning everyday and professional tasks. Each task is an end-to-end workflow that takes human users a median of about 1.6 hours, scored with a binary-completion metric, and targets challenges such as dynamic environments, cross-source reasoning, and implicit-state inference.

  72.6%

  Self-reported

• ARC-AGI

  #1 / 11

  The Abstraction and Reasoning Corpus for Artificial General Intelligence (ARC-AGI) is a benchmark designed to test general intelligence and abstract reasoning capabilities through visual grid-based transformation tasks. Each task consists of 2-5 demonstration pairs showing input grids transformed into output grids according to underlying rules, with test-takers required to infer these rules and apply them to novel test inputs. The benchmark uses colored grids (up to 30x30) with 10 discrete colors/symbols, designed to measure human-like general fluid intelligence and skill-acquisition efficiency with minimal prior knowledge.

  98.5%

  Self-reported

• Humanity's Last Exam (with tools, text-only)

  #3 / 10

  Text-only Humanity's Last Exam variant evaluated with tool use enabled.

  57.2%

  Self-reported

• ExploitBench

  #1 / 8

  ExploitBench is a cybersecurity benchmark that evaluates a model's ability to discover and exploit software vulnerabilities, reported as the fraction of challenges where the model captures the target (Cap%).

  100.0%

  Self-reported

• ExploitGym

  #1 / 8

  ExploitGym is a large-scale, realistic benchmark built from real-world vulnerabilities across userspace programs, Google's V8 engine, and the Linux kernel. Given a vulnerability and a proof-of-vulnerability input, agents must craft a working end-to-end exploit that achieves unauthorized code execution.

  42.4%

  Self-reported

• BenchCAD (with Python tool)

  #3 / 7

  BenchCAD variant evaluated with access to a Python tool for programmatic CAD reasoning.

  95.9%

  Self-reported

• SEC-bench Pro

  #1 / 7

  SEC-bench Pro is a self-evolving software-security benchmark that measures agent bug hunting on critical, high-complexity systems. It instantiates validated vulnerabilities across the V8 and SpiderMonkey JavaScript engines as reproducible vulnerability-discovery and proof-of-concept-generation tasks with oracle-based validation.

  85.4%

  Self-reported

• Terminal-Bench-Science 0.1

  #1 / 7

  Terminal-Bench-Science 0.1 is a Stanford-led community benchmark of 70 tasks drawn from scientific research workflows across the life, physical, earth, mathematical, and engineering sciences. Tasks are authored and reviewed by scientists and researchers; scores are typically reported as accuracy with relatively large standard error due to strongly bimodal per-task outcomes.

  64.6%

  Self-reported

• ARC-AGI-3

  #1 / 5

  ARC-AGI-3 is the third-generation Abstraction and Reasoning Corpus benchmark, an interactive-reasoning evaluation designed to measure fluid, novel problem-solving ability that remains far from saturated for frontier models.

  99.9%

  Self-reported

• FrontierMath Tier 4 (v2)

  #2 / 5

  FrontierMath Tier 4 subset from the v2 evaluation release.

  97.6%

  Self-reported

• LifeSciBench

  #1 / 5

  LifeSciBench is an expert-authored, expert-reviewed benchmark of 750 open-ended life-science research tasks spanning seven workflows and seven biological domains. Responses are graded against 19,020 physician- and scientist-written rubric criteria rather than multiple-choice answers, and most tasks require interpreting attached artifacts such as figures, PDFs, and sequence files.

  60.3%

  Self-reported

• GeneBench-Pro

  #1 / 4

  GeneBench-Pro is a research-level benchmark of 129 multi-stage computational-biology problems spanning genomics, quantitative biology, and translational biomedicine. Each problem gives the agent a messy dataset, brief context, and a target estimand, and requires navigating dependent inferential decision points to reach a verifiable answer.

  37.8%

  Self-reported