Benchmark Almanac

FoodTruck Bench

Gemini 4 Carbon $77,218 leads since October 2026
Not in model cards

Agents run a food truck in Austin, Texas for a 30-day simulation, choosing locations, menus, prices, inventory, staff and loans with 34 tools. Humans can play the same game.

60 models tested on six locations. Scored by business results at the end of the run.

Kind
Simulated companies
Level
Everyday
Scored
Settled by events
Answers
Agent task
Tasks
All public
Strongest model tested
Gemini 4 Carbon #2 on the capability index
Last model added
November 2026
Scores refreshed
November 12th, 2026
63
Entries on its own leaderboard, 59 models we track

Gemini 4 Carbon leads with $77,218

Best score per entry from the benchmark's own leaderboard, read on November 12th, 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.

63 entries. Scroll sideways for the rest.

  1. $77,218 Gemini 4 Carbon
  2. $75,264 Claude Opus 5
  3. $67,598 Tencent Hy4
  4. $63,464 MiniMax-M3.1-Flash
  5. $62,732 Claude Fable 5.1
  6. $61,408 GPT-5.5
  7. $59,537 Grok 4.6
  8. $57,761 GPT-6 Astra
  9. $53,289 DeepSeek V4.1 Flash 1017
  10. $53,229 GPT-5.6 Sol
  11. $50,179 DeepSeek V4.1 Flash
  12. $49,519 Claude Opus 4.6
  13. $49,373 MiMo-V2.6 Pro
  14. $44,443 Qwen 3.8 Max
  15. $42,561 Grok 4.7
  16. $41,904 Muse Spark 1.2
  17. $39,037 Tencent Hy4 preview
  18. $37,350 DeepSeek V4 Pro 0813
  19. $34,639 Muse Ember
  20. $34,586 Grok 4.5
  21. $32,932 GPT-5.6 Luna
  22. $28,081 GPT-5.2
  23. $28,055 Tencent Hy3
  24. $27,880 Grok 4.3
  25. $27,827 Qwen3.7-Max
  26. $27,168 Qwen 3.8 27B
  27. $27,142 DeepSeek-V4-Pro
  28. $24,923 Nex N2 Pro
  29. $24,878 Gemma 4 31B IT
  30. $24,039 DeepSeek V4 Flash 0731
  31. $22,404 Kimi K3
  32. $22,388 MiMo-V2.5-Pro
  33. $22,311 Gemini 3.5 Flash
  34. $17,426 Claude Sonnet 4.6
  35. $17,211 GPT-6 Luna
  36. $17,199 Gemini 3 Pro
  37. $15,317 Qwen 3.6 35B-A3B
  38. $13,985 Solar Pro 4
  39. $13,914 MiMo-V2.6 Flash
  40. $12,736 Gemini 3.1 Pro CT
  41. $10,425 Muse Glimmer
  42. $7,668 Qwen 3.6 Plus
  43. $5,504 DeepSeek-V4-Flash
  44. $4,386 Gemma 4 26B A4B
  45. $3,517 Nemotron 3 Ultra
  46. $2,058 DeepSeek-V3.2
  47. $1,388 Claude Sonnet 4.5
  48. $1,338 Grok 4.20
  49. $1,245 GLM-5.3-Flash
  50. $962 Nemotron 3 Super
  51. $882 MiniMax-M3
  52. $817 Grok 4.1 Fast
  53. $598 MiMo-V2-Omni
  54. $470 GPT-5.4 Mini
  55. $166 Claude Haiku 4.5
  56. $92 gpt-oss-120b
  57. $50 GPT-5 mini
  58. $30 Kimi K2.5
  59. $-210 GLM-5
  60. $-218 Qwen 3.5 397B
  61. $-317 MiniMax-M2.5
  62. $-525 Qwen3 VL 235B
  63. $-679 Qwen3.5-9B
  1. 1 Gemini 4 Carbon median run; 30 days $77,218
  2. 2 Claude Opus 5 median run; 30 days $75,264
  3. 3 Tencent Hy4 median run; 30 days $67,598
  4. 4 MiniMax-M3.1-Flash median run; 30 days $63,464
  5. 5 Claude Fable 5.1 median run; 30 days $62,732
  6. 6 GPT-5.5 median run; 30 days $61,408
  7. 7 Grok 4.6 median run; 30 days $59,537
  8. 8 GPT-6 Astra median run; 30 days $57,761
  9. 9 DeepSeek V4.1 Flash 1017 median run; 30 days $53,289
  10. 10 GPT-5.6 Sol median run; 30 days $53,229
  11. 11 DeepSeek V4.1 Flash median run; 30 days $50,179
  12. 12 Claude Opus 4.6 median run; 30 days $49,519
  13. 13 MiMo-V2.6 Pro median run; 30 days $49,373
  14. 14 Qwen 3.8 Max median run; 30 days $44,443
  15. 15 Grok 4.7 median run; 30 days $42,561
  16. 16 Muse Spark 1.2 median run; 30 days $41,904
  17. 17 Tencent Hy4 preview median run; 30 days $39,037
  18. 18 DeepSeek V4 Pro 0813 median run; 30 days $37,350
  19. 19 Muse Ember median run; 30 days $34,639
  20. 20 Grok 4.5 median run; 30 days $34,586
  21. 21 GPT-5.6 Luna median run; 30 days $32,932
  22. 22 GPT-5.2 median run; 30 days $28,081
  23. 23 Tencent Hy3 median run; 30 days $28,055
  24. 24 Grok 4.3 median run; 30 days $27,880
  25. 25 Qwen3.7-Max median run; 30 days $27,827
  26. 26 Qwen 3.8 27B median run; 30 days; NVFP4 $27,168
  27. 27 DeepSeek-V4-Pro median run; 30 days $27,142
  28. 28 Nex N2 Pro median run; 30 days $24,923
  29. 29 Gemma 4 31B IT median run; 30 days $24,878
  30. 30 DeepSeek V4 Flash 0731 median run; 30 days $24,039
  31. 31 Kimi K3 median run; 30 days $22,404
  32. 32 MiMo-V2.5-Pro median run; 30 days $22,388
  33. 33 Gemini 3.5 Flash median run; 30 days $22,311
  34. 34 Claude Sonnet 4.6 median run; 30 days $17,426
  35. 35 GPT-6 Luna median run; 30 days $17,211
  36. 36 Gemini 3 Pro median run; 30 days $17,199
  37. 37 Qwen 3.6 35B-A3B median run; 30 days $15,317
  38. 38 Solar Pro 4 median run; 30 days $13,985
  39. 39 MiMo-V2.6 Flash median run; 30 days $13,914
  40. 40 Gemini 3.1 Pro CT median run; 30 days $12,736
  41. 41 Muse Glimmer median run; 30 days $10,425
  42. 42 Qwen 3.6 Plus median run; 30 days $7,668
  43. 43 DeepSeek-V4-Flash median run; 30 days $5,504
  44. 44 Gemma 4 26B A4B median run; 30 days $4,386
  45. 45 Nemotron 3 Ultra median run; 30 days $3,517
  46. 46 DeepSeek-V3.2 median run; 30 days $2,058
  47. 47 Claude Sonnet 4.5 median run; 30 days $1,388
  48. 48 Grok 4.20 median run; 30 days $1,338
  49. 49 GLM-5.3-Flash median run; 30 days $1,245
  50. 50 Nemotron 3 Super median run; 30 days $962
  51. 51 MiniMax-M3 median run; 30 days $882
  52. 52 Grok 4.1 Fast median run; 30 days $817
  53. 53 MiMo-V2-Omni median run; 30 days $598
  54. 54 GPT-5.4 Mini median run; 30 days $470
  55. 55 Claude Haiku 4.5 median run; 30 days $166
  56. 56 gpt-oss-120b median run; 30 days $92
  57. 57 GPT-5 mini median run; 30 days $50
  58. 58 Kimi K2.5 median run; 30 days $30
  59. 59 GLM-5 median run; 30 days $-210
  60. 60 Qwen 3.5 397B median run; 30 days $-218
  61. 61 MiniMax-M2.5 median run; 30 days $-317
  62. 62 Qwen3 VL 235B median run; 30 days $-525
  63. 63 Qwen3.5-9B median run; 30 days $-679

Source: Official FoodTruck Bench leaderboard

#EntryScoreSettingDateSource
1 Gemini 4 Carbon $77,218 median run; 30 days Oct 2026 Official leaderboard
2 Claude Opus 5 $75,264 median run; 30 days Jul 2026 Official leaderboard
3 Tencent Hy4 $67,598 median run; 30 days Nov 2026 Official leaderboard
4 MiniMax-M3.1-Flash $63,464 median run; 30 days Oct 2026 Official leaderboard
5 Claude Fable 5.1 $62,732 median run; 30 days Sep 2026 Official leaderboard
6 GPT-5.5 $61,408 median run; 30 days Apr 2026 Official leaderboard
7 Grok 4.6 $59,537 median run; 30 days Aug 2026 Official leaderboard
8 GPT-6 Astra $57,761 median run; 30 days Sep 2026 Official leaderboard
9 DeepSeek V4.1 Flash 1017 $53,289 median run; 30 days Oct 2026 Official leaderboard
10 GPT-5.6 Sol $53,229 median run; 30 days Jul 2026 Official leaderboard
11 DeepSeek V4.1 Flash $50,179 median run; 30 days Sep 2026 Official leaderboard
12 Claude Opus 4.6 $49,519 median run; 30 days Feb 2026 Official leaderboard
13 MiMo-V2.6 Pro $49,373 median run; 30 days Sep 2026 Official leaderboard
14 Qwen 3.8 Max $44,443 median run; 30 days Aug 2026 Official leaderboard
15 Grok 4.7 $42,561 median run; 30 days Sep 2026 Official leaderboard
16 Muse Spark 1.2 $41,904 median run; 30 days Aug 2026 Official leaderboard
17 Tencent Hy4 preview $39,037 median run; 30 days Aug 2026 Official leaderboard
18 DeepSeek V4 Pro 0813 $37,350 median run; 30 days Aug 2026 Official leaderboard
19 Muse Ember $34,639 median run; 30 days Nov 2026 Official leaderboard
20 Grok 4.5 $34,586 median run; 30 days Jul 2026 Official leaderboard
21 GPT-5.6 Luna $32,932 median run; 30 days Jul 2026 Official leaderboard
22 GPT-5.2 $28,081 median run; 30 days Dec 2025 Official leaderboard
23 Tencent Hy3 $28,055 median run; 30 days Jul 2026 Official leaderboard
24 Grok 4.3 $27,880 median run; 30 days Apr 2026 Official leaderboard
25 Qwen3.7-Max $27,827 median run; 30 days May 2026 Official leaderboard
26 Qwen 3.8 27B $27,168 median run; 30 days; NVFP4 Aug 2026 Official leaderboard
27 DeepSeek-V4-Pro $27,142 median run; 30 days Apr 2026 Official leaderboard
28 Nex N2 Pro $24,923 median run; 30 days Official leaderboard
29 Gemma 4 31B IT $24,878 median run; 30 days Apr 2026 Official leaderboard
30 DeepSeek V4 Flash 0731 $24,039 median run; 30 days Jul 2026 Official leaderboard
31 Kimi K3 $22,404 median run; 30 days Jul 2026 Official leaderboard
32 MiMo-V2.5-Pro $22,388 median run; 30 days Apr 2026 Official leaderboard
33 Gemini 3.5 Flash $22,311 median run; 30 days May 2026 Official leaderboard
34 Claude Sonnet 4.6 $17,426 median run; 30 days Feb 2026 Official leaderboard
35 GPT-6 Luna $17,211 median run; 30 days Sep 2026 Official leaderboard
36 Gemini 3 Pro $17,199 median run; 30 days Nov 2025 Official leaderboard
37 Qwen 3.6 35B-A3B $15,317 median run; 30 days Apr 2026 Official leaderboard
38 Solar Pro 4 $13,985 median run; 30 days Aug 2026 Official leaderboard
39 MiMo-V2.6 Flash $13,914 median run; 30 days Sep 2026 Official leaderboard
40 Gemini 3.1 Pro CT $12,736 median run; 30 days Official leaderboard
41 Muse Glimmer $10,425 median run; 30 days Aug 2026 Official leaderboard
42 Qwen 3.6 Plus $7,668 median run; 30 days Mar 2026 Official leaderboard
43 DeepSeek-V4-Flash $5,504 median run; 30 days Apr 2026 Official leaderboard
44 Gemma 4 26B A4B $4,386 median run; 30 days Apr 2026 Official leaderboard
45 Nemotron 3 Ultra $3,517 median run; 30 days Jun 2026 Official leaderboard
46 DeepSeek-V3.2 $2,058 median run; 30 days Dec 2025 Official leaderboard
47 Claude Sonnet 4.5 $1,388 median run; 30 days Sep 2025 Official leaderboard
48 Grok 4.20 $1,338 median run; 30 days Feb 2026 Official leaderboard
49 GLM-5.3-Flash $1,245 median run; 30 days Aug 2026 Official leaderboard
50 Nemotron 3 Super $962 median run; 30 days Mar 2026 Official leaderboard
51 MiniMax-M3 $882 median run; 30 days Jun 2026 Official leaderboard
52 Grok 4.1 Fast $817 median run; 30 days Nov 2025 Official leaderboard
53 MiMo-V2-Omni $598 median run; 30 days Mar 2026 Official leaderboard
54 GPT-5.4 Mini $470 median run; 30 days Mar 2026 Official leaderboard
55 Claude Haiku 4.5 $166 median run; 30 days Oct 2025 Official leaderboard
56 gpt-oss-120b $92 median run; 30 days Aug 2025 Official leaderboard
57 GPT-5 mini $50 median run; 30 days Aug 2025 Official leaderboard
58 Kimi K2.5 $30 median run; 30 days Jan 2026 Official leaderboard
59 GLM-5 $-210 median run; 30 days Feb 2026 Official leaderboard
60 Qwen 3.5 397B $-218 median run; 30 days Official leaderboard
61 MiniMax-M2.5 $-317 median run; 30 days Feb 2026 Official leaderboard
62 Qwen3 VL 235B $-525 median run; 30 days Official leaderboard
63 Qwen3.5-9B $-679 median run; 30 days Feb 2026 Official leaderboard

8 models have held number one on the official leaderboard since 2023

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

OpenAI Anthropic Google DeepMind
Own leaderboard 8 leaders
Gemini 4 Carbon $77,218

Source: Official FoodTruck Bench leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard
$-20,000 $0 $20,000 $40,000 $60,000 $80,000 2026 2027 $-20,000 $0 $20,000 $40,000 $60,000 $80,000 2026 2027

Source: Official FoodTruck Bench leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Economic work

All 97 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
  • 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%
  • OfficeQA In use Grounded reasoning over a large corpus of U.S. Treasury Bulletin documents: locate tables and compute precise numeric answers. Claude Opus 4.7 86.3%