Benchmark Almanac

DRBench

GPT-5 63.8 leads since September 2025
Not in model cards

Agents investigate open-ended business questions across enterprise applications and produce reports. The evaluator checks whether the reports recover relevant information and support their claims.

Released
September 2025
Kind
Deep research
Level
University
Scored
Graded by a model
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Actions, Text
Interaction
Agentic
Environment
APIs and tools, Browser
Refresh
One fixed set
Run by
Anyone
Strongest model tested
GPT-5 #71 on the capability index
Last model added
August 2025
Scores refreshed
November 20th, 2026
9
Entries in its paper, 5 models we track

GPT-5 leads with 63.8 among models released by August 2025 GPT-5 leads with 63.8 among models released by August 2025

Best score per entry 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. 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.

The newest model with a score here came out in August 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-5 No planning 63.8
  2. 2 DeepSeek-V3.1 Complex research planning 55.0
  3. 3 Qwen2.5-72B Complex research planning 49.5
  4. 4 GPT-5 mini Simple research planning 48.9
  5. 5 Llama-3.1-405B-Instruct Complex research planning 43.7
  6. 6 GPT-4o (snapshot unspecified) Simple research planning 43.6
  7. 7 Llama-3.1-70b-Instruct No planning 38.8
  8. 8 gpt-oss-120b No planning 35.4
  9. 9 GPT-4o-mini (snapshot unspecified) Simple research planning 31.4

Source: The DRBench paper

#EntryScoreSettingDateSource
1 GPT-5 63.8 No planning, DRBA Aug 2025 Official leaderboard
2 DeepSeek-V3.1 55.0 Complex research planning, DRBA Aug 2025 Official leaderboard
3 Qwen2.5-72B 49.5 Complex research planning, DRBA Sep 2024 Official leaderboard
4 GPT-5 mini 48.9 Simple research planning, DRBA Aug 2025 Official leaderboard
5 Llama-3.1-405B-Instruct 43.7 Complex research planning, DRBA Official leaderboard
6 GPT-4o (snapshot unspecified) 43.6 Simple research planning, DRBA Official leaderboard
7 Llama-3.1-70b-Instruct 38.8 No planning, DRBA Official leaderboard
8 gpt-oss-120b 35.4 No planning, DRBA Aug 2025 Official leaderboard
9 GPT-4o-mini (snapshot unspecified) 31.4 Simple research planning, DRBA Official leaderboard
#ModelScoreSettingDateSource
1 GPT-5 63.8 No planning, DRBA Aug 2025 Official leaderboard
2 DeepSeek-V3.1 55.0 Complex research planning, DRBA Aug 2025 Official leaderboard
3 Qwen2.5-72B 49.5 Complex research planning, DRBA Sep 2024 Official leaderboard
4 GPT-5 mini 48.9 Simple research planning, DRBA Aug 2025 Official leaderboard
5 gpt-oss-120b 35.4 No planning, DRBA Aug 2025 Official leaderboard

Rankings by part

We read 4 parts from the benchmark’s paper, each ranking models on one part of the test, such as Insight Recall, Factuality and Distractor Avoidance. GPT-5 leads 3 of them, and Qwen2.5-72B leads the other one. Arrows show how far a model moved from its overall rank among the models we track.

Insight Recall

5 models we track have a score here.

  1. 1 GPT-5 37.5
  2. 2 DeepSeek-V3.1 28.2
  3. 3 GPT-5 mini ↑1 24.6
  4. 4 Qwen2.5-72B ↓1 24.4
  5. 5 gpt-oss-120b 19.3

Scores at each effort level

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.

Lowest effort to highest, light to dark
GPT-5 DRBA
Complex res…
63.8
DeepSeek-V3.1 DRBA
No planning Complex res…
55.0
Qwen2.5-72B DRBA
Simple rese… Complex res…
49.5
GPT-5 mini DRBA
Complex res… Simple rese…
48.9
gpt-oss-120b DRBA
Simple rese… Complex res…
35.4

Source: The DRBench paper

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5 No planning DRBA 63.8
GPT-5 Simple research planning DRBA 62.6
GPT-5 Complex research planning DRBA 62.0
DeepSeek-V3.1 Complex research planning DRBA 55.0
DeepSeek-V3.1 Simple research planning DRBA 49.9
Qwen2.5-72B Complex research planning DRBA 49.5
DeepSeek-V3.1 No planning DRBA 49.2
Qwen2.5-72B No planning DRBA 49.0
GPT-5 mini Simple research planning DRBA 48.9
GPT-5 mini No planning DRBA 46.5
Qwen2.5-72B Simple research planning DRBA 46.3
GPT-5 mini Complex research planning DRBA 45.7
gpt-oss-120b No planning DRBA 35.4
gpt-oss-120b Complex research planning DRBA 34.7
gpt-oss-120b Simple research planning DRBA 30.9

Scores at each effort level

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.

Lowest effort to highest, light to dark
GPT-5 DRBA
Complex res…
63.8
DeepSeek-V3.1 DRBA
No planning Complex res…
55.0
Qwen2.5-72B DRBA
Simple rese… Complex res…
49.5
GPT-5 mini DRBA
Complex res… Simple rese…
48.9
gpt-oss-120b DRBA
Simple rese… Complex res…
35.4

Source: The DRBench paper

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5 No planning DRBA 63.8
GPT-5 Simple research planning DRBA 62.6
GPT-5 Complex research planning DRBA 62.0
DeepSeek-V3.1 Complex research planning DRBA 55.0
DeepSeek-V3.1 Simple research planning DRBA 49.9
Qwen2.5-72B Complex research planning DRBA 49.5
DeepSeek-V3.1 No planning DRBA 49.2
Qwen2.5-72B No planning DRBA 49.0
GPT-5 mini Simple research planning DRBA 48.9
GPT-5 mini No planning DRBA 46.5
Qwen2.5-72B Simple research planning DRBA 46.3
GPT-5 mini Complex research planning DRBA 45.7
gpt-oss-120b No planning DRBA 35.4
gpt-oss-120b Complex research planning DRBA 34.7
gpt-oss-120b Simple research planning DRBA 30.9

Number one over time

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

OpenAI
Benchmark paper Not tested since Sep 2025
GPT-5 63.8

Source: The DRBench paper

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
In the benchmark’s paper
30.0 40.0 50.0 60.0 70.0 2025 30.0 40.0 50.0 60.0 70.0 2025

Source: Official DRBench 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 Tool use

All 265 in Tool use
  • AutomationBench In use Agent completes end-to-end business workflows inside a simulated company exposing dozens of REST APIs. Graded on resulting state. Gemini 4 Neon 56.3%
  • MCP Atlas In use Multi-step workflows over real MCP servers. Scores tool-use competency on realistic tasks. Gemini 4 Argon 90.7%
  • BFCL V4 In use Models call functions and tools, including multi-turn agentic tasks with web search and memory. Checked against expected calls and outcomes. Qwen3.7-Max 75%
  • τ³-bench Banking In use Agents act as banking customer-service reps over many turns with a simulated user, following policy. Pass^1 on the final database state. Qwen 3.8 Max 55.1%
  • Toolathlon-Verified In use 108 long-horizon tasks across 32 apps through about 600 MCP tools. Execution-based checkers inspect final environment state. GLM-5.3-Flash 78.4%
  • DeepSearchQA In use Questions that need multi-step search and synthesis. Final answers graded against references. Kimi K3 95%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 15 months ago, in August 2025, for DeepSeek-V3.1. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the benchmark’s paper (9 entries).
  • Pass Models scored 5 models we track, plus 4 other entries on its leaderboard.
  • Warning Newest score The newest came 15 months ago, in August 2025, for DeepSeek-V3.1.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5 (#71).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 15 rows for models we track and 4 other entries 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