Benchmark Almanac

CRM-Agent-Bench

Gemini 3.8 Flash 83.6% leads since October 2026
Not in model cards

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
Language
English
In and out
Text → Actions
Interaction
Agentic
Items from
Generated
Refresh
One fixed set
Run by
Anyone
Strongest model tested
Claude Opus 5 #11 on the capability index
Last model added
September 2026
Scores refreshed
November 20th, 2026
16
Models scored on its own leaderboard

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:

  1. 1 Gemini 3.8 Flash 83.6%
  2. 2 Claude Opus 5 82.7%
  3. 3 Claude Sonnet 4.6 82%
  4. 4 GPT-5.6 Terra 81.6%
  5. 5 GPT-5.5 81.3%
  6. 6 Claude Sonnet 5 81.3%
  7. 7 GPT-5.6 Sol 81.3%
  8. 8 GPT-5.6 Luna 80.7%
  9. 9 Inkling-Small 79.6%
  10. 10 Gemini 3.1 Pro 79.1%
  11. 11 GPT-5.4 Nano 79.1%
  12. 12 Muse Glimmer 78.7%
  13. 13 Claude Haiku 4.5 76%
  14. 14 Nemotron 3 Super 65.6%
  15. 15 GPT-4.1 nano 64.3%
  16. 16 Nemotron 3.5 Lightning 30B-A3B 64.3%

Source: Official CRM-Agent-Bench leaderboard

#ModelScoreSettingDateSource
1 Gemini 3.8 Flash 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 Inkling-Small 79.6% Jul 2026 Official leaderboard
10 Gemini 3.1 Pro 79.1% Feb 2026 Official leaderboard
11 GPT-5.4 Nano 79.1% Mar 2026 Official leaderboard
12 Muse Glimmer 78.7% Aug 2026 Official leaderboard
13 Claude Haiku 4.5 76% Oct 2025 Official leaderboard
14 Nemotron 3 Super 65.6% Mar 2026 Official leaderboard
15 GPT-4.1 nano 64.3% Apr 2025 Official leaderboard
16 Nemotron 3.5 Lightning 30B-A3B 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. 1 Gemini 3.8 Flash 91.6%
  2. 2 Claude Opus 5 88.4%
  3. 3 Gemini 3.1 Pro ↑7 87.4%
  4. 4 GPT-5.4 Nano ↑7 86.3%
  5. 5 Claude Sonnet 5 ↑1 86.3%
  6. 6 GPT-5.6 Terra ↓2 85.3%
  7. 7 Claude Haiku 4.5 ↑6 84.2%
  8. 8 GPT-5.6 Luna 84.2%
  9. 9 Inkling-Small 84.2%
  10. 10 Claude Sonnet 4.6 ↓7 82.1%

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.

OpenAI Google DeepMind Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $1.00 $10 $100 $1,000 Cost of the full run, log scale Gemini 3.8 Flash GPT-5.6 Luna GPT-4.1 nano 0% 20% 40% 60% 80% 100% $1.00 $10 $100 $1,000 Cost of the full run, log scale Gemini 3.8 Flash GPT-4.1 nano

Source: Official CRM-Agent-Bench leaderboard

ModelEffortHarnessScoreCost of the full runOutput tokens
Gemini 3.8 Flash 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
Inkling-Small 79.6%
Gemini 3.1 Pro 79.1% $62
GPT-5.4 Nano 79.1% $5.86
Muse Glimmer 78.7%
Claude Haiku 4.5 76% $41
Nemotron 3 Super 65.6% $8.60
GPT-4.1 nano 64.3% $2.14
Nemotron 3.5 Lightning 30B-A3B 64.3%

Number one over time

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

Google DeepMind
Own leaderboard Not tested since Oct 2026

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.

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%) Gemini 3.8 Flash Claude Sonnet 4.6 Claude Haiku 4.5 GPT-4.1 nano 0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) Gemini 3.8 Flash Claude Haiku 4.5

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

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

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.

How we check