CRM-Agent-Bench
Agents handle customer and business requests in a simulated enterprise. Executable checks verify tool arguments, action order, and which records changed across 555 tasks.
- Made by
- Salesforce
- Released
- October 2026
- Kind
- Running a business
- Level
- University
- Scored
- Checked automatically
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Subject
- Economics and business
- Language
- English
- Interaction
- Agentic
- Environment
- APIs and tools, A simulator
- Items from
- Generated
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Agents, Robotics, Law and finance
- Strongest model tested
- Claude Opus 5 #11 on the capability index
- Last model added
- September 2026
- Scores refreshed
- November 20th, 2026
Gemini 3.8 Flash leads with 83.6% 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. Every run is in the next chart.
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:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 13 more
- 83.6
Gemini 3.8 Flash
- 82.7 Claude Opus 5
- 82.0 Claude Sonnet 4.6
- 81.6 GPT-5.6 Terra
- 81.3 GPT-5.5
- 81.3 Claude Sonnet 5
- 81.3 GPT-5.6 Sol
- 80.7 GPT-5.6 Luna
- 79.6
Inkling-Small
- 79.1
Gemini 3.1 Pro
- 79.1 GPT-5.4 Nano
- 78.7
Muse Glimmer
- 76.0 Claude Haiku 4.5
- 65.6
Nemotron 3 Super
- 64.3 GPT-4.1 nano
- 64.3
Nemotron 3.5 Lightning 30B-A3B
- 1
Gemini 3.8 Flash 83.6%
- 2 Claude Opus 5 82.7%
- 3 Claude Sonnet 4.6 82%
- 4 GPT-5.6 Terra 81.6%
- 5 GPT-5.5 81.3%
- 6 Claude Sonnet 5 81.3%
- 7 GPT-5.6 Sol 81.3%
- 8 GPT-5.6 Luna 80.7%
- 9
Inkling-Small 79.6%
- 10
Gemini 3.1 Pro 79.1%
- 11 GPT-5.4 Nano 79.1%
- 12
Muse Glimmer 78.7%
- 13 Claude Haiku 4.5 76%
- 14
Nemotron 3 Super 65.6%
- 15 GPT-4.1 nano 64.3%
- 16
Nemotron 3.5 Lightning 30B-A3B 64.3%
Source: Official CRM-Agent-Bench leaderboard
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 83.6% | Sep 2026 | Official leaderboard | |
| 2 | Claude Opus 5 | 82.7% | Jul 2026 | Official leaderboard | |
| 3 | Claude Sonnet 4.6 | 82% | Feb 2026 | Official leaderboard | |
| 4 | GPT-5.6 Terra | 81.6% | Jul 2026 | Official leaderboard | |
| 5 | GPT-5.5 | 81.3% | Apr 2026 | Official leaderboard | |
| 6 | Claude Sonnet 5 | 81.3% | Jun 2026 | Official leaderboard | |
| 7 | GPT-5.6 Sol | 81.3% | Jul 2026 | Official leaderboard | |
| 8 | GPT-5.6 Luna | 80.7% | Jul 2026 | Official leaderboard | |
| 9 | | 79.6% | Jul 2026 | Official leaderboard | |
| 10 | | 79.1% | Feb 2026 | Official leaderboard | |
| 11 | GPT-5.4 Nano | 79.1% | Mar 2026 | Official leaderboard | |
| 12 | | 78.7% | Aug 2026 | Official leaderboard | |
| 13 | Claude Haiku 4.5 | 76% | Oct 2025 | Official leaderboard | |
| 14 | | 65.6% | Mar 2026 | Official leaderboard | |
| 15 | GPT-4.1 nano | 64.3% | Apr 2025 | Official leaderboard | |
| 16 | | 64.3% | Aug 2026 | Official leaderboard |
Rankings by part
We read 9 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Retail, Customer support and Field service. Claude Sonnet 4.6 leads 3 of them, and 6 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
Retail
16 models we track have a score here.
- 1
Gemini 3.8 Flash 91.6%
- 2 Claude Opus 5 88.4%
- 3
Gemini 3.1 Pro ↑7 87.4%
- 4 GPT-5.4 Nano ↑7 86.3%
- 5 Claude Sonnet 5 ↑1 86.3%
- 6 GPT-5.6 Terra ↓2 85.3%
- 7 Claude Haiku 4.5 ↑6 84.2%
- 8 GPT-5.6 Luna 84.2%
- 9
Inkling-Small 84.2%
- 10 Claude Sonnet 4.6 ↓7 82.1%
Customer support
16 models we track have a score here.
- 1
Muse Glimmer ↑11 68%
- 2 Claude Opus 5 65.6%
- 3
Gemini 3.8 Flash ↓2 63.2%
- 4 Claude Sonnet 4.6 ↓1 62.4%
- 5 Claude Sonnet 5 ↑1 61.6%
- 6 GPT-5.6 Sol ↑1 61.6%
- 7 GPT-5.6 Terra ↓3 61.6%
- 8 GPT-5.5 ↓3 60%
- 9 GPT-5.6 Luna ↓1 60%
- 10
Inkling-Small ↓1 56.8%
Field service
16 models we track have a score here.
- 1 Claude Sonnet 4.6 ↑2 100%
- 2 GPT-5.6 Sol ↑5 100%
- 3
Gemini 3.1 Pro ↑7 95.7%
- 4 GPT-5.5 ↑1 95.7%
- 5 Claude Sonnet 5 ↑1 95.7%
- 6 GPT-5.6 Luna ↑2 95.7%
- 7 GPT-5.6 Terra ↓3 95.7%
- 8 Claude Haiku 4.5 ↑5 91.3%
- 9 Claude Opus 5 ↓7 91.3%
- 10
Muse Glimmer ↑2 91.3%
IT service management
16 models we track have a score here.
- 1 GPT-5.6 Luna ↑7 95.5%
- 2 GPT-5.6 Terra ↑2 93.9%
- 3
Gemini 3.1 Pro ↑7 90.9%
- 4
Muse Glimmer ↑8 89.4%
- 5
Gemini 3.8 Flash ↓4 89.4%
- 6 Claude Sonnet 4.6 ↓3 87.9%
- 7 GPT-5.6 Sol 87.9%
- 8 GPT-5.5 ↓3 86.4%
- 9 Claude Opus 5 ↓7 86.4%
- 10 GPT-5.4 Nano ↑1 84.8%
Payments
16 models we track have a score here.
- 1 GPT-5.4 Nano ↑10 95.9%
- 2 GPT-5.5 ↑3 95.9%
- 3 GPT-5.6 Sol ↑4 95.9%
- 4 GPT-5.6 Luna ↑4 94.5%
- 5
Gemini 3.8 Flash ↓4 94.5%
- 6 GPT-5.6 Terra ↓2 93.2%
- 7 Claude Opus 5 ↓5 93.2%
- 8 Claude Sonnet 4.6 ↓5 91.8%
- 9
Inkling-Small 89%
- 10 Claude Sonnet 5 ↓4 87.7%
Travel
16 models we track have a score here.
- 1
Inkling-Small ↑8 83.9%
- 2 Claude Sonnet 4.6 ↑1 81.7%
- 3 GPT-5.5 ↑2 81.7%
- 4
Gemini 3.8 Flash ↓3 81.7%
- 5 Claude Sonnet 5 ↑1 80.6%
- 6 GPT-5.6 Sol ↑1 79.6%
- 7 Claude Opus 5 ↓5 78.5%
- 8
Gemini 3.1 Pro ↑2 77.4%
- 9
Muse Glimmer ↑3 77.4%
- 10 GPT-5.4 Nano ↑1 75.3%
Onboarding
16 models we track have a score here.
- 1 Claude Sonnet 4.6 ↑2 95.1%
- 2
Gemini 3.1 Pro ↑8 95.1%
- 3 Claude Sonnet 5 ↑3 95.1%
- 4
Nemotron 3 Super ↑10 92.7%
- 5 GPT-5.5 92.7%
- 6 GPT-5.6 Luna ↑2 92.7%
- 7 GPT-5.6 Terra ↓3 92.7%
- 8
Inkling-Small ↑1 92.7%
- 9 Claude Opus 5 ↓7 92.7%
- 10 Claude Haiku 4.5 ↑3 90.2%
Sales
16 models we track have a score here.
- 1 GPT-5.6 Sol ↑6 100%
- 2 GPT-5.6 Terra ↑2 94.4%
- 3
Gemini 3.8 Flash ↓2 94.4%
- 4 GPT-5.4 Nano ↑7 88.9%
- 5 GPT-5.5 88.9%
- 6 Claude Sonnet 5 88.9%
- 7 GPT-5.6 Luna ↑1 88.9%
- 8 Claude Opus 5 ↓6 88.9%
- 9
Muse Glimmer ↑3 88.9%
- 10 GPT-4.1 nano ↑5 83.3%
Marketing
16 models we track have a score here.
- 1 Claude Sonnet 4.6 ↑2 100%
- 2 Claude Sonnet 5 ↑4 100%
- 3 GPT-5.4 Nano ↑8 95.2%
- 4 GPT-5.5 ↑1 95.2%
- 5 Claude Opus 5 ↓3 95.2%
- 6 Claude Haiku 4.5 ↑7 90.5%
- 7
Inkling-Small ↑2 90.5%
- 8
Muse Glimmer ↑4 90.5%
- 9
Gemini 3.8 Flash ↓8 90.5%
- 10 GPT-4.1 nano ↑5 85.7%
The best score, 83.6%, costs $16 for the full run
Runs on the cost frontier score higher than any run that's cheaper. GPT-5.6 Luna comes within 2.9 points of the best for $4.77 for the full run, 3 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.
Source: Official CRM-Agent-Bench leaderboard
| Model | Effort | Harness | Score | Cost of the full run | Output tokens |
|---|---|---|---|---|---|
| 83.6% | $16 | ||||
| Claude Opus 5 | 82.7% | $191 | |||
| Claude Sonnet 4.6 | 82% | $71 | |||
| GPT-5.6 Terra | 81.6% | $52 | |||
| GPT-5.5 | 81.3% | $93 | |||
| Claude Sonnet 5 | 81.3% | $72 | |||
| GPT-5.6 Sol | 81.3% | $107 | |||
| GPT-5.6 Luna | 80.7% | $4.77 | |||
| 79.6% | |||||
| 79.1% | $62 | ||||
| GPT-5.4 Nano | 79.1% | $5.86 | |||
| 78.7% | |||||
| Claude Haiku 4.5 | 76% | $41 | |||
| 65.6% | $8.60 | ||||
| GPT-4.1 nano | 64.3% | $2.14 | |||
| 64.3% |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official CRM-Agent-Bench leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official CRM-Agent-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
- 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
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 (16 entries).
- Pass Models scored 16 models we track.
- Pass Newest score The newest is from September 2026, for Gemini 3.8 Flash.
- Note Frontier models None of the top 10 has a score. The strongest tested is Claude Opus 5 (#11).
- Pass Headroom The best is 83.6%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on crm-agent-bench.github.io: 16 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.