Benchmark Almanac

AscendGYM

Claude Opus 5 74.0 leads since November 2026
Not in model cards

Agents optimize existing kernels for Ascend processors using tensor shapes from real workloads. The benchmark measures speedup and token consumption.

The paper describes 48 kernels at three difficulty levels. It reports a speedup score adjusted for token cost.

Made by
AscendGYM team
Released
November 2026
Kind
GPU kernels
Level
Expert
Scored
Checked automatically
Answers
Code
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Code → Code
Interaction
Agentic
Environment
Terminal
Items from
Real work and use
Refresh
One fixed set
Run by
Anyone
In views
Agents
Strongest model tested
Claude Opus 5 #11 on the capability index
Last model added
August 2026
Scores refreshed
November 20th, 2026
7
Models scored on its own leaderboard

Claude Opus 5 leads with 74.0 among models released by August 2026

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best.

The newest model with a score here came out in August 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Claude Opus 5 74.0
  2. 2 GPT-5.6 Terra 58.4
  3. 3 GLM-5.3 44.0
  4. 4 GPT-5.6 Luna Default 34.7
  5. 5 DeepSeek-V4-Flash 11.5
  6. 6 Qwen3.5-122B-A10B 7.90
  7. 7 Qwen 3.6 35B-A3B 7.16

Source: Official AscendGYM leaderboard

#ModelScoreSettingDateSource
1 Claude Opus 5 74.0 claude-code Jul 2026 Official leaderboard
2 GPT-5.6 Terra 58.4 codex Jul 2026 Official leaderboard
3 GLM-5.3 44.0 claude-code Aug 2026 Official leaderboard
4 GPT-5.6 Luna 34.7 Default, default-warm Jul 2026 Official leaderboard
5 DeepSeek-V4-Flash 11.5 codex Apr 2026 Official leaderboard
6 Qwen3.5-122B-A10B 7.90 opencode Feb 2026 Official leaderboard
7 Qwen 3.6 35B-A3B 7.16 opencode Apr 2026 Official leaderboard

Number one over time

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

Anthropic
Own leaderboard 1 leader

Source: Official AscendGYM leaderboard

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
On the benchmark’s own leaderboard
0.00 20.0 40.0 60.0 80.0 2026 0.00 20.0 40.0 60.0 80.0 2026

Source: Official AscendGYM leaderboard

No lab model card reports it yet

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

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 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (7 entries).
  • Pass Models scored 7 models we track.
  • Pass Newest score The newest is from August 2026, for GLM-5.3.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude Opus 5 (#11).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on slai-aitp.github.io: 21 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
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