Benchmark Almanac

PMPP-Hard

Kimi K3 0.71 leads since July 2026
Inkling 2 Small 494.3 leads since November 2026
In use

The official leaderboard hasn’t scored a model released after DeepSeek V4 Flash 0731 (July 2026). Labs report their own numbers for 2 newer models, such as Inkling 2 Small at 494.3.

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.

Thinking Machines reports an adapted version with task and grading fixes.

Released
July 2026
Kind
GPU kernels
Level
Research frontier
Scored
Checked automatically
Answers
Agent task
Measures
Capability
Language
English
In and out
Text, Code → Code, Actions
Interaction
Agentic
Environment
Terminal
Refresh
One fixed set
Run by
Anyone
In views
Agents
Strongest model tested
GPT-5.6 Sol #14 on the capability index
Last in a launch post
November 2026
Last model added
November 2026
Scores refreshed
November 18th, 2026
11
Entries on its own leaderboard, 10 models we track
2
Launch posts list it, from 1 lab
1
System card reports it, from 1 lab

Kimi K3 leads with 0.71 among models released by July 2026 Kimi K3 leads with 0.71 among models released by July 2026

Best score per entry from the benchmark's own leaderboard, read on November 18th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

Inkling 2 Small leads with 494.3, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model. Striped bars are claims, and dotted bars are reruns by Epoch AI or another board.

The newest model with a score here came out in July 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 Kimi K3 max; 4,500 s 0.71
  2. 2 Claude Opus 4.8 medium; 1,500 s 0.62
  3. 3 GPT-5.6 Sol xhigh; 1,500 s 0.59
  4. 4 GPT-5.5 medium; 1,500 s 0.52
  5. 5 Claude Sonnet 5 high; 1,500 s 0.52
  6. 6 GLM-5.2 medium; 1,500 s 0.39
  7. 7 DeepSeek-V4-Pro medium; 1,500 s 0.35
  8. 8 DeepSeek V4 Flash 0731 high; 4,500 s 0.33
  9. 9 GPT-5.6 Terra medium; 1,500 s 0.32
  10. 10 Nemotron Ultra medium; 1,500 s 0.26
  11. 11 GPT-5.6 Luna medium; 1,500 s 0.23

Source: Official PMPP-Hard leaderboard

#EntryScoreSettingDateSource
1 Kimi K3 0.71 max; 4,500 s, kimi_code Jul 2026 Official leaderboard
2 Claude Opus 4.8 0.62 medium; 1,500 s, Codex May 2026 Official leaderboard
3 GPT-5.6 Sol 0.59 xhigh; 1,500 s, Codex Jul 2026 Official leaderboard
4 GPT-5.5 0.52 medium; 1,500 s, Codex Apr 2026 Official leaderboard
5 Claude Sonnet 5 0.52 high; 1,500 s, Claude Code Jun 2026 Official leaderboard
6 GLM-5.2 0.39 medium; 1,500 s, Codex Jun 2026 Official leaderboard
7 DeepSeek-V4-Pro 0.35 medium; 1,500 s, Codex Apr 2026 Official leaderboard
8 DeepSeek V4 Flash 0731 0.33 high; 4,500 s, Codex Jul 2026 Official leaderboard
9 GPT-5.6 Terra 0.32 medium; 1,500 s, Codex Jul 2026 Official leaderboard
10 Nemotron Ultra 0.26 medium; 1,500 s, Codex Official leaderboard
11 GPT-5.6 Luna 0.23 medium; 1,500 s, Codex Jul 2026 Official leaderboard
#ModelScoreSettingDateSource
1 Kimi K3 0.71 max; 4,500 s, kimi_code Jul 2026 Official leaderboard
2 Claude Opus 4.8 0.62 medium; 1,500 s, Codex May 2026 Official leaderboard
3 GPT-5.6 Sol 0.59 xhigh; 1,500 s, Codex Jul 2026 Official leaderboard
4 GPT-5.5 0.52 medium; 1,500 s, Codex Apr 2026 Official leaderboard
5 Claude Sonnet 5 0.52 high; 1,500 s, Claude Code Jun 2026 Official leaderboard
6 GLM-5.2 0.39 medium; 1,500 s, Codex Jun 2026 Official leaderboard
7 DeepSeek-V4-Pro 0.35 medium; 1,500 s, Codex Apr 2026 Official leaderboard
8 DeepSeek V4 Flash 0731 0.33 high; 4,500 s, Codex Jul 2026 Official leaderboard
9 GPT-5.6 Terra 0.32 medium; 1,500 s, Codex Jul 2026 Official leaderboard
10 GPT-5.6 Luna 0.23 medium; 1,500 s, Codex Jul 2026 Official leaderboard
#ModelScoreSettingDateSource
1 Inkling 2 Small 494.3 Nov 2026 Lab model card
2 Inkling 2 Mini 455.8 Nov 2026 Lab model card
3 Kimi K3 0.71 max; 4,500 s, kimi_code Jul 2026 Official leaderboard
4 Claude Opus 4.8 0.62 medium; 1,500 s, Codex May 2026 Official leaderboard
5 GPT-5.6 Sol 0.59 xhigh; 1,500 s, Codex Jul 2026 Official leaderboard
6 GPT-5.5 0.52 medium; 1,500 s, Codex Apr 2026 Official leaderboard
7 Claude Sonnet 5 0.52 high; 1,500 s, Claude Code Jun 2026 Official leaderboard
8 GLM-5.2 0.39 medium; 1,500 s, Codex Jun 2026 Official leaderboard
9 DeepSeek-V4-Pro 0.35 medium; 1,500 s, Codex Apr 2026 Official leaderboard
10 DeepSeek V4 Flash 0731 0.33 high; 4,500 s, Codex Jul 2026 Official leaderboard
11 GPT-5.6 Terra 0.32 medium; 1,500 s, Codex Jul 2026 Official leaderboard
12 GPT-5.6 Luna 0.23 medium; 1,500 s, Codex Jul 2026 Official leaderboard

Labs report scores for 2 models in their cards. Switch the scores above to rank them together.

2 models have held number one on the official leaderboard since its release in July 2026 3 models have held number one with claims included since its release in July 2026

Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for the benchmark’s own leaderboard and what labs claim in their cards.

Anthropic Moonshot AI Thinking Machines
Own leaderboard Not tested since Jul 2026
Kimi K3 0.71
Lab claims 1 leader

Sources: Official PMPP-Hard leaderboard and 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
On the benchmark’s own leaderboard Reported by the lab in its model card
0.00 100.0 200.0 300.0 400.0 500.0 2026 2027 0.00 100.0 200.0 300.0 400.0 500.0 2026 2027

Sources: Official PMPP-Hard leaderboard and labs’ model cards

One lab put it in a launch post in the last six months

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

Thinking Machines
DateLaunch postScoreAs the post writes it
Nov 20, 2026 Inkling 2 Mini 455.8 PMPP-Hard (AUC, 0 to 10h), pi v0.83.0
Nov 5, 2026 Inkling 2 Small 494.3 PMPP-Hard AUC, 494.27 task-hours

One system card reports it

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.

DateDocumentScoreAs the document writes itSection
Nov 20, 2026 Inkling 2 Mini model card 455.8 PMPP-Hard (AUC, 0 to 10h), pi v0.83.0 Capabilities

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%
  • KernelBench In use Replace PyTorch operators with faster custom GPU kernels. Scored on correctness and speedup. MiniMax-M3 28.8%
  • OpenAI PRs Only in system cards Agents replicate internal OpenAI pull requests from a pre-PR branch and a prompt. Graded by hidden unit tests as pass@1. GPT-5.2 55%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Task access check. We don’t know whether the tasks are public. Every benchmark

Catalog record 1 warning 1 pass
  • Warning Task access We don’t know whether the tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (11 entries), launch posts (2 scores) and system cards (1 score).
  • Pass Models scored 12 models we track, plus 1 other entry on its leaderboard.
  • Pass Newest score The newest is from November 2026, for Inkling 2 Mini.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.6 Sol (#14).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 10 rows for models we track and 1 other entry at the last read.
  • Pass Last read November 18th, 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