Benchmark Almanac

SDR-Bench

Claude Sonnet 4.6 55.8% leads since February 2026
Not in model cards

Models personalize sales pitches using customer research. A judge compares each pitch with points from successful human outreach.

The October 2026 paper reports Claude Sonnet 4.6 first at 55.8% weighted coverage, ahead of GPT models, STORM and ODR.

Released
February 2026
Kind
Running a business
Level
University
Scored
Graded by a model
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Text, Actions
Interaction
Agentic
Environment
Browser
Items from
Real work and use
Refresh
One fixed set
Run by
Anyone
Occupations
Sales
Strongest model tested
GPT-5.4 #28 on the capability index
Last model added
March 2026
Scores refreshed
November 20th, 2026
8
Entries on its own leaderboard, 6 models we track

Claude Sonnet 4.6 leads with 55.8% among models released by March 2026 Claude Sonnet 4.6 leads with 55.8% among models released by March 2026

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. 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 March 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 Sonnet 4.6 Search 55.8%
  2. 2 GPT-5.4 Mini Search 44.6%
  3. 3 GPT-5.4 Search 44.3%
  4. 4 STORM-QWEN-2.5 Search 42.5%
  5. 5 GPT-4o Mar 2025, Direct 38.1%
  6. 6 GPT-4o mini Search 37.5%
  7. 7 Qwen2.5-72B Search 36.8%
  8. 8 ODR-QWEN-2.5 Search 33.5%

Source: Official SDR-Bench leaderboard

#EntryScoreSettingDateSource
1 Claude Sonnet 4.6 55.8% Search Feb 2026 Official leaderboard
2 GPT-5.4 Mini 44.6% Search Mar 2026 Official leaderboard
3 GPT-5.4 44.3% Search Mar 2026 Official leaderboard
4 STORM-QWEN-2.5 42.5% Search Official leaderboard
5 GPT-4o Mar 2025 38.1% Direct Mar 2025 Official leaderboard
6 GPT-4o mini 37.5% Search Jul 2024 Official leaderboard
7 Qwen2.5-72B 36.8% Search Sep 2024 Official leaderboard
8 ODR-QWEN-2.5 33.5% Search Official leaderboard
#ModelScoreSettingDateSource
1 Claude Sonnet 4.6 55.8% Search Feb 2026 Official leaderboard
2 GPT-5.4 Mini 44.6% Search Mar 2026 Official leaderboard
3 GPT-5.4 44.3% Search Mar 2026 Official leaderboard
4 GPT-4o Mar 2025 38.1% Direct Mar 2025 Official leaderboard
5 GPT-4o mini 37.5% Search Jul 2024 Official leaderboard
6 Qwen2.5-72B 36.8% Search Sep 2024 Official leaderboard

Rankings by part

We read 4 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Tech., Mfg. and Energy. Claude Sonnet 4.6 leads every one of them. Arrows show how far a model moved from its overall rank among the models we track.

Tech.

6 models we track have a score here.

  1. 1 Claude Sonnet 4.6 52.6%
  2. 2 GPT-5.4 Mini 42.6%
  3. 3 GPT-5.4 39.7%
  4. 4 GPT-4o Mar 2025 35.5%
  5. 5 GPT-4o mini 33.3%
  6. 6 Qwen2.5-72B 32.1%

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
Claude Sonnet 4.6
Direct Search
55.8%
GPT-5.4 Mini
Direct
44.6%
GPT-5.4
Direct
44.3%
GPT-4o
Search Direct
38.1%
GPT-4o mini
Direct Search
37.5%
Qwen2.5-72B
Direct
36.8%

Source: Official SDR-Bench leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Sonnet 4.6 Search 55.8%
Claude Sonnet 4.6 Direct 49.9%
GPT-5.4 Mini Search 44.6%
GPT-5.4 Search 44.3%
GPT-5.4 Mini Direct 44.1%
GPT-5.4 Direct 44%
GPT-4o Mar 2025 Direct 38.1%
GPT-4o mini Search 37.5%
Qwen2.5-72B Search 36.8%
Qwen2.5-72B Direct 35.7%
GPT-4o Mar 2025 Search 35.4%
GPT-4o mini Direct 33.9%

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
Claude Sonnet 4.6
Direct Search
55.8%
GPT-5.4 Mini
Direct
44.6%
GPT-5.4
Direct
44.3%
GPT-4o
Search Direct
38.1%
GPT-4o mini
Direct Search
37.5%
Qwen2.5-72B
Direct
36.8%

Source: Official SDR-Bench leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Sonnet 4.6 Search 55.8%
Claude Sonnet 4.6 Direct 49.9%
GPT-5.4 Mini Search 44.6%
GPT-5.4 Search 44.3%
GPT-5.4 Mini Direct 44.1%
GPT-5.4 Direct 44%
GPT-4o Mar 2025 Direct 38.1%
GPT-4o mini Search 37.5%
Qwen2.5-72B Search 36.8%
Qwen2.5-72B Direct 35.7%
GPT-4o Mar 2025 Search 35.4%
GPT-4o mini Direct 33.9%

2 models have held number one on the official leaderboard since its release in February 2026

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

OpenAI Anthropic
Own leaderboard Not tested since Mar 2026

Source: Official SDR-Bench 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 Best so far
0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) Claude Sonnet 4.6 GPT-4o GPT-4o mini 0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) Claude Sonnet 4.6 GPT-4o

Source: Official SDR-Bench 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 Economic work

All 210 in Economic work
  • JobBench In use 130 tasks from 35 occupations workers most want delegated. Agent works through messy reference files, graded by fact-anchored binary rubrics. GLM-5.3 61.4%
  • GDPval In use Real work tasks from 44 occupations with reference files. Experts blindly compare model deliverables with expert work. Metric is win-or-tie rate. Seed2.1 Pro 87.9%
  • GDPval-AA v2.1 In use Agents with shell and web access complete 220 GDPval gold tasks from 44 occupations. Blind pairwise comparisons give Elo ratings. Claude Opus 5.5 1,846
  • AA-Briefcase v1.1 In use Version 1.1 of Artificial Analysis's multi-hour office work evaluation, with Elo ratings fitted by Crowd-BT. Claude Opus 5.5 1,822
  • Agents' Last Exam v1.1 In use Agents complete 122 public long-horizon professional workflow tasks across 55 sub-industries. Deterministic graders check outcomes against hidden references. GPT-6.2 Astra 47.5%
  • CoWorkBench In use Agent completes long-horizon productivity tasks across computer science, finance, law and medicine. Qwen in-house scoring (method unpublished). Claude Fable 5 75.9%

Data audit

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 8 months ago, in March 2026, for GPT-5.4 Mini. 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 official leaderboard (8 entries).
  • Pass Models scored 6 models we track, plus 2 other entries on its leaderboard.
  • Warning Newest score The newest came 8 months ago, in March 2026, for GPT-5.4 Mini.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.4 (#28).
  • Pass Headroom The best is 55.8%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on behavior-in-the-wild.github.io: 12 rows for models we track and 2 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