Benchmark Almanac

Infrathon

Claude Opus 5.5 91.2 leads since November 2026
Not in model cards

Agents optimize AI infrastructure under correctness and fidelity constraints. Hidden workloads measure performance gains, with additional checks for rule compliance.

The paper evaluates six models over 150 runs with a 24-hour budget. Metrics depend on the task and should retain their workload and resource limits.

Released
November 2026
Kind
Speed and optimization
Level
Expert
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Efficiency
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-6 Astra #4 on the capability index
Last model added
September 2026
Scores refreshed
November 20th, 2026

In the authors’ words

In this work, we study whether frontier agent can continually improve real-world AI infrastructure over long horizons. We introduce Infrathon, a benchmark of diverse, long-horizon AI infrastructure optimization tasks across different levels. We evaluate six models across 150 runs with 24-hour budgets, totaling over 10,000 GPU-hours. Our results show that model rankings vary with the evaluation horizon, while 135 of the 150 runs achieve further improvements during the second half of their budgets. These findings indicate that shorter evaluations can overlook both optimization gains and changes in relative model performance. We also find agents can discover solutions that outperform human-designed baselines: Claude Opus 5.5 increases the mean KV-cache compression ratio from 6.82% with the human-designed Quant-VideoGen codec to 19.86%. Analysis of agent trajectories identifies task violations, unreliable testing, and premature completion as barriers to sustained improvement. Together, Our Infrathon framework provide resources for constructing long-horizon tasks, evaluating frontier agents, and discovering good infrastructure solutions.

Abstract of Infrathon: Benchmarking Language Agents for Real-World, Long-Horizon AI Infra Optimization (arXiv:2611.11995, CC0)
6
Models scored on its own leaderboard

Claude Opus 5.5 leads with 91.2 among models released by September 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 September 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 5.5 91.2
  2. 2 GPT-6 Astra 75.2
  3. 3 GPT-6.1 Sol 54.4
  4. 4 Qwen3.8 Flash 40.8
  5. 5 DeepSeek V4.1 Flash 30.4
  6. 6 GLM-5.3-Flash 8.00

Source: Official Infrathon leaderboard

#ModelScore ±SettingDateSource
1 Claude Opus 5.5 91.2 85.6 to 96 Sep 2026 Official leaderboard
2 GPT-6 Astra 75.2 65.6 to 83.2 Sep 2026 Official leaderboard
3 GPT-6.1 Sol 54.4 44.8 to 63.2 Sep 2026 Official leaderboard
4 Qwen3.8 Flash 40.8 ±8.8 Aug 2026 Official leaderboard
5 DeepSeek V4.1 Flash 30.4 23.2 to 38.4 Sep 2026 Official leaderboard
6 GLM-5.3-Flash 8.00 1.6 to 16 Aug 2026 Official leaderboard

Rankings by part

We read 25 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Attn-QAT NVFP4 attention, K-grouped FP8 Wgrad and Flux/Comet MoE primitive. Claude Opus 5.5 leads 17 of them, and 3 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Attn-QAT NVFP4 attention

TFLOPS. 12 hours. 1×H100 6 models we track have a score here.

  1. 1 Claude Opus 5.5 370.4
  2. 2 GPT-6 Astra 345.2
  3. 3 GPT-6.1 Sol 321.6
  4. 4 DeepSeek V4.1 Flash ↑1 282.7
  5. 5 Qwen3.8 Flash ↓1 254.9
  6. 6 GLM-5.3-Flash 110.3

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 Infrathon 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 80.0 100.0 0.00 20.0 40.0 60.0 80.0 100.0

Source: Official Infrathon 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 Coding

All 523 in Coding
  • SWE-bench Verified In use 500 human-validated GitHub issues from 12 Python repos. Agent writes a patch that must pass hidden tests. Claude Opus 4.7 83.5%
  • LiveCodeBench v6 In use Competitive programming problems collected after model cutoffs from LeetCode, AtCoder, Codeforces. Scored pass@1 on hidden tests. DeepSeek-V4-Pro 93.5%
  • SWE-bench Multilingual In use 300 GitHub-issue tasks across nine programming languages in SWE-bench format. Patch graded by repository tests. Claude Fable 5.2 96.6%
  • DeepSWE In use 113 original long-horizon engineering tasks across 91 repos and five languages, written from scratch. Hand-written verifiers accept alternative solutions. GPT-6 Astra 74.1%
  • SciCode In use Scientist-curated research coding problems split into subproblems across natural sciences. Code checked against test cases. Seed2.1 Pro 59.8%
  • Codeforces rating In use Model solves recent Codeforces contest problems. An Elo-style rating is computed from simulated contest performance. DeepSeek V4.1 Flash 3,471

Data audit

All 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (6 entries).
  • Pass Models scored 6 models we track.
  • Pass Newest score The newest is from September 2026, for GPT-6.1 Sol.
  • Pass Frontier models 3 of the top 10 have a score, led by GPT-6 Astra (#4).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on infrathonx.github.io: 6 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