Benchmark Almanac

Internal Research Debugging Evaluation

GPT-6.2 Astra 81.8% leads since November 2026
Only in system cards

Agents debug difficult problems from OpenAI research work that take researchers hours or days. Scored as mean rubric score.

In-house OpenAI AI self-improvement evaluation. The tasks are not public.

Made by
OpenAI
Released
April 2026
Kind
Labs' internal R&D tests
Level
Research frontier
Scored
Graded by a model
Answers
Agent task
Tasks
Kept private
Measures
Capability
Language
English
In and out
Text, Code → Code, Actions
Interaction
Agentic
Environment
Terminal
Items from
Real work and use
Refresh
One fixed set
Strongest model tested
GPT-6.2 Astra #1 on the capability index
Last in a system card
November 2026
Last model added
November 2026
8
Models scored in lab cards
14
System cards report it, from 1 lab

GPT-6.2 Astra leads with 81.8%

Best score per model as labs report them in their own cards, drawn striped. Labs pick their own settings, so these compare less cleanly than independent runs.

  1. 1 GPT-6.2 Astra 81.8%
  2. 2 GPT-6 Astra 78%
  3. 3 GPT-6.1 Terra 76.4%
  4. 4 GPT-6.1 Sol 75.5%
  5. 5 GPT-6.2 Sol 74.7%
  6. 6 GPT-6 Sol 64.2%
  7. 7 GPT-5.5 50.5%
  8. 8 GPT-6 Luna 46.6%

Source: Labs’ model cards

#ModelScoreSettingDateSource
1 GPT-6.2 Astra 81.8% Nov 2026 Lab model card
2 GPT-6 Astra 78% Sep 2026 Lab model card
3 GPT-6.1 Terra 76.4% Oct 2026 Lab model card
4 GPT-6.1 Sol 75.5% Sep 2026 Lab model card
5 GPT-6.2 Sol 74.7% Nov 2026 Lab model card
6 GPT-6 Sol 64.2% Sep 2026 Lab model card
7 GPT-5.5 50.5% Apr 2026 Lab model card
8 GPT-6 Luna 46.6% Sep 2026 Lab model card

3 models have held number one in labs’ own numbers since its release in April 2026

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Lab claims 3 leaders

Source: Labs’ model cards

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) GPT-6.2 Astra GPT-6 Astra GPT-5.5 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) GPT-6.2 Astra GPT-5.5

Source: Labs’ model cards

Labs report it only in system cards

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark. Some system cards do, below.

14 system cards report it, from 1 lab

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests. OpenAI accounts for 11 of the 11 releases that named it since 2024.

DateDocumentScoreAs the document writes itSection
Nov 17, 2026 GPT-6.2 Sol system card 74.7 Internal Research Debugging, mean rubric score (Table 5) AI R&D risk
Nov 17, 2026 GPT-6.2 Astra system card 81.8 Internal Research Debugging, mean rubric score (Table 5) AI R&D risk
Oct 26, 2026 GPT-6.1 Terra system card addendum 76.4 Internal Research Debugging Evaluation, mean rubric score (Figure 29, 9.0.3.1) AI R&D risk
Sep 29, 2026 GPT-6.1 Sol system card addendum 75.5 Internal Research Debugging Evaluation, mean rubric score (9.1.3.1, Figure 37) AI R&D risk
Sep 22, 2026 GPT-6 Sol system card 64.2 Internal Research Debugging, mean rubric reward, printed chart label (Figure 84, A.8.1.3.1) AI R&D risk
Sep 22, 2026 GPT-6 Luna system card 46.6 Internal Research Debugging, mean rubric reward, printed chart label (Figure 84, A.8.1.3.1) AI R&D risk
Sep 3, 2026 GPT-6 Astra system card 78 Internal Research Debugging Evaluation, score (10.1.3.1) AI R&D risk
Jul 9, 2026 GPT-5.6 Terra system card Internal Research Debugging Evaluation, rubric score by cost and latency, chart without printed numbers (Figures 37, 38, 9.1.3.1) AI R&D risk
Jul 9, 2026 GPT-5.6 Sol system card Internal Research Debugging Evaluation, rubric score by cost and latency, chart without printed numbers (Figures 37, 38, 9.1.3.1) AI R&D risk
Jul 9, 2026 GPT-5.6 Luna system card Internal Research Debugging Evaluation, rubric score by cost and latency, chart without printed numbers (Figures 37, 38, 9.1.3.1) AI R&D risk
Jun 25, 2026 GPT-5.6 Terra system card Internal Research Debugging Evaluation, rubric score by cost and latency, chart without printed numbers (Figures 37, 38, 9.1.3.1) AI R&D risk
Jun 25, 2026 GPT-5.6 Sol system card Internal Research Debugging Evaluation, rubric score by cost and latency, chart without printed numbers (Figures 37, 38, 9.1.3.1) AI R&D risk
Jun 25, 2026 GPT-5.6 Luna system card Internal Research Debugging Evaluation, rubric score by cost and latency, chart without printed numbers (Figures 37, 38, 9.1.3.1) AI R&D risk
Apr 23, 2026 GPT-5.5 system card 50.5 Internal Research Debugging Eval, mean rubric score, printed chart label (Figure 20, 9.1.3.3) AI R&D risk

More in AI research

All 164 in AI research
  • PaperBench In use Agent replicates ICML 2024 papers from scratch. LLM judge grades against author-written hierarchical rubrics. Qwen 3.8 Max 93%
  • MLS-Bench In use Agent must invent one algorithmic change in an ML research scaffold whose gain transfers across settings. Scored relative to human baselines. Claude Fable 5.1 50.3%
  • PostTrainBench 1.2 In use CLI agent gets a base model, one H100 and a time budget to post-train it for a target benchmark. Scored on normalized gain. Claude Fable 5.1 44.6%
  • MLE-bench In use Kaggle-style ML engineering competitions. Agent trains models and is scored against competition leaderboards. GPT-6.2 Astra 96.1%
  • PMPP-Hard In use Agents write GPU kernels for 69 tasks that must be correct, and on 48 of them also run within 1.25 times the reference time. The score is the area under the solved-tasks curve over a 10-hour budget, out of 690 task-hours. Kimi K3 0.71
  • KernelBench In use Replace PyTorch operators with faster custom GPU kernels. Scored on correctness and speedup. MiniMax-M3 28.8%

Data audit

All 7 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access The tasks are private.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Note Independent runs Only labs’ own numbers, as expected for an in-house eval. Scores from system cards (8 scores).
  • Pass Models scored 8 models we track.
  • Pass Newest score The newest is from November 2026, for GPT-6.2 Astra.
  • Pass Frontier models 4 of the top 10 have a score, led by GPT-6.2 Astra (#1).
  • Pass Headroom The best is 81.8%, under 85%, so it still separates models.
Leaderboard import 1 note
  • Note Importer No public leaderboard to read.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check