TREC Million LLMs 2025
Systems rank simulated expert language models for a new question using their answers to earlier questions. Each expert has access to a different document cluster.
SRCB tested Qwen3 models. The organizers report scores on 16 questions with relevant answers, using GPT-5 judgments against reference answers.
- Made by
- NIST TREC Million LLMs Track
- Released
- November 2024
- Kind
- Questions that need search
- Level
- University
- Scored
- Checked automatically
- Answers
- Short answer
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Single turn
- Items from
- Generated
- Refresh
- One fixed set
- Run by
- Anyone
submission_4 tops the leaderboard with 0.20
Best score per entry from the benchmark's own leaderboard, read on November 20th, 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.
- 0.20 submission_4
- 0.17 infolab_UD_run2
- 0.17 infolab_UD_run1
- 0.16 ensemble_v3
- 0.15 ensemble_v2
- 0.14 ensemble_v1
- 0.14 non-neural
- 0.14 infolab_UD_run4
- 0.13 r-ef-rm3-1000-ss
- 0.13 r-f-rm3-1000-rr-ss
- 0.12 submission_5
- 0.12 r-f-rm3-1000-ss
- 0.10 infolab_UD_run5
- 0.10 q-f-rm3-100-ss
- 0.10 infolab_UD_run3
- 0.09 llm-f-100
- 0.09 lightgbm_job266431
- 0.09 mllm-indelab-09-17
- 0.04 lambdamart_profiles
- 1 submission_4 0.20
- 2 infolab_UD_run2 0.17
- 3 infolab_UD_run1 0.17
- 4 ensemble_v3 0.16
- 5 ensemble_v2 0.15
- 6 ensemble_v1 0.14
- 7 non-neural 0.14
- 8 infolab_UD_run4 0.14
- 9 r-ef-rm3-1000-ss 0.13
- 10 r-f-rm3-1000-rr-ss 0.13
- 11 submission_5 0.12
- 12 r-f-rm3-1000-ss 0.12
- 13 infolab_UD_run5 0.10
- 14 q-f-rm3-100-ss 0.10
- 15 infolab_UD_run3 0.10
- 16 llm-f-100 0.09
- 17 lightgbm_job266431 0.09
- 18 mllm-indelab-09-17 0.09
- 19 lambdamart_profiles 0.04
Source: Official TREC Million LLMs 2025 leaderboard
| # | Entry | Score | Setting | Submitted by | Date | Link | Source |
|---|---|---|---|---|---|---|---|
| 1 | submission_4 | 0.20 | SRCB | trec.nist.gov | Official leaderboard | ||
| 2 | infolab_UD_run2 | 0.17 | UDInfo | trec.nist.gov | Official leaderboard | ||
| 3 | infolab_UD_run1 | 0.17 | UDInfo | trec.nist.gov | Official leaderboard | ||
| 4 | ensemble_v3 | 0.16 | SRCB | trec.nist.gov | Official leaderboard | ||
| 5 | ensemble_v2 | 0.15 | SRCB | trec.nist.gov | Official leaderboard | ||
| 6 | ensemble_v1 | 0.14 | SRCB | trec.nist.gov | Official leaderboard | ||
| 7 | non-neural | 0.14 | indelab | trec.nist.gov | Official leaderboard | ||
| 8 | infolab_UD_run4 | 0.14 | UDInfo | trec.nist.gov | Official leaderboard | ||
| 9 | r-ef-rm3-1000-ss | 0.13 | GenAIus | trec.nist.gov | Official leaderboard | ||
| 10 | r-f-rm3-1000-rr-ss | 0.13 | GenAIus | trec.nist.gov | Official leaderboard | ||
| 11 | submission_5 | 0.12 | SRCB | trec.nist.gov | Official leaderboard | ||
| 12 | r-f-rm3-1000-ss | 0.12 | GenAIus | trec.nist.gov | Official leaderboard | ||
| 13 | infolab_UD_run5 | 0.10 | UDInfo | trec.nist.gov | Official leaderboard | ||
| 14 | q-f-rm3-100-ss | 0.10 | GenAIus | trec.nist.gov | Official leaderboard | ||
| 15 | infolab_UD_run3 | 0.10 | UDInfo | trec.nist.gov | Official leaderboard | ||
| 16 | llm-f-100 | 0.09 | GenAIus | trec.nist.gov | Official leaderboard | ||
| 17 | lightgbm_job266431 | 0.09 | uvairlab | trec.nist.gov | Official leaderboard | ||
| 18 | mllm-indelab-09-17 | 0.09 | indelab | trec.nist.gov | Official leaderboard | ||
| 19 | lambdamart_profiles | 0.04 | paulphys | trec.nist.gov | Official 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 Knowledge
- MMLU-Pro In use Harder 10-option version of MMLU with reasoning-focused questions across 14 subjects.
Intern-S2-397B 89.8%
- MMLU In use 57-subject multiple-choice exam questions from elementary to professional level. Scored by accuracy.
Seed1.8 92.3%
- SuperGPQA In use 26,529 graduate-level multiple-choice questions across 285 disciplines. Accuracy.
Qwen 3.6 Max (Preview) 73.9%
- WideSearch In use 200 bilingual broad info-seeking tasks: agent collects many atomic facts into a table. Scored by success rate and row/item F1.
Tencent Hy4 preview 83.9%
- BBEH micro average In use Models solve harder replacements for each of the 23 BIG-Bench Hard tasks, 4,520 examples in total. The score is the share of all examples answered correctly.
Gemma 4 31B IT 74.4%
- HLE-Diamond In use Models answer 1,000 cleaned expert questions from Humanity's Last Exam, 500 on reasoning and 500 on knowledge, closed book. Graded on exact answers. Claude Fable 5.2 77.9%
Data audit
All 8 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
- Pass Independent runs Scores from the official leaderboard (19 entries).
- Pass Models scored 19 entries on its leaderboard, none a model we track.
- Note Newest score No scores yet.
- Note Frontier models No models scored yet.
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on trec.nist.gov: 19 entries at the last read, none a model we track.
- 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.