TRACE
Models identify reward hacking in agent trajectories from coding environments.
The paper reports model results.
- Made by
- Patronus AI
- Released
- January 2026
- Kind
- Fixing real repositories
- Level
- Expert
- Scored
- Checked automatically
- Answers
- Short answer
- Tasks
- All public
- Measures
- Capability
- Risk
- Loss of control
- Language
- English
- Interaction
- Single turn
- Items from
- Generated
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Loss of control
- Strongest model tested
- GPT-5.2 #51 on the capability index
- Last model added
- December 2025
- Scores refreshed
- November 20th, 2026
GPT-5.2 leads with 63% among models released by December 2025
Best score per model from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.
The newest model with a score here came out in December 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
Source: The TRACE paper
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.2 | 63% | Contrastive, N = 10, high reasoning | Dec 2025 | Official leaderboard |
| 2 | Kimi K2 Thinking | 62% | Contrastive, N = 10 | Nov 2025 | Official leaderboard |
| 3 | | 61% | Contrastive, N = 10 | Nov 2025 | Official leaderboard |
| 4 | Claude Opus 4.5 | 59% | Contrastive, N = 10 | Nov 2025 | Official leaderboard |
| 5 | GLM-4.7 | 59% | Contrastive, N = 10 | Dec 2025 | Official leaderboard |
| 6 | | 58% | Contrastive, N = 10 | Dec 2025 | Official leaderboard |
More effort adds a median 18 points
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: The TRACE paper
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-5.2 | Contrastive, N = 10, high reasoning | 63% | |||
| Kimi K2 Thinking | Contrastive, N = 10 | 62% | |||
| Contrastive, N = 10 | 61% | ||||
| Claude Opus 4.5 | Contrastive, N = 10 | 59% | |||
| GLM-4.7 | Contrastive, N = 10 | 59% | |||
| Contrastive, N = 10 | 58% | ||||
| GPT-5.2 | Isolated, N = 1, high reasoning | 45% | |||
| Kimi K2 Thinking | Isolated, N = 1 | 37% | |||
| Isolated, N = 1 | 35% | ||||
| Isolated, N = 1 | 35% | ||||
| GLM-4.7 | Isolated, N = 1 | 34% | |||
| Claude Opus 4.5 | Isolated, N = 1 | 25% |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The TRACE paper
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official TRACE 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
- 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
9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 11 months ago, in December 2025, for GLM-4.7. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
- Pass Independent runs Scores from the benchmark’s paper (6 entries).
- Pass Models scored 6 models we track.
- Warning Newest score The newest came 11 months ago, in December 2025, for GLM-4.7.
- Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.2 (#51).
- Pass Headroom The best is 63%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on proceedings.mlr.press: 12 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.