1
Claude Fable 5.2 Anthropic Provisional · 2 of 5 groups
88
84 – 96 – –
$16.00
10 Nov 2026
Spring Prompt benches 84CatalogBench 81 · Reliably publish-ready 71.4%; Sales brief: reliably publish-ready 44.6% ROASBench 100 · Overall score 61.3 BulletBench 68 · Ladder Elo 491 (95% range 230–714) Knowledge work 96DeckBench 91 · Deck rating 1,403 (95% range 1,311–1,512) SurveyBench 100 · Analysis rating 1,670 (95% range 1,580–1,808)
No results: Agents and tool use, Writing and preference, Factuality
2
GPT-6.2 Astra OpenAI Provisional · 2 of 5 groups
86
90 – 75 – –
$16.00
18 Nov 2026
Spring Prompt benches 90CatalogBench 92 · Reliably publish-ready 73.2%; Sales brief: reliably publish-ready 69.6% ROASBench 94 · Overall score 57.6 BulletBench 79 · Ladder Elo 557 (95% range 370–727) Knowledge work 75DeckBench 54 · Deck rating 1,188 (95% range 1,082–1,318) SurveyBench 96 · Analysis rating 1,476 (95% range 1,402–1,580)
No results: Agents and tool use, Writing and preference, Factuality
3
GPT-6.2 Sol OpenAI Provisional · 2 of 5 groups
84
85 – 81 – –
$4.00
17 Nov 2026
Spring Prompt benches 85CatalogBench 97 · Reliably publish-ready 75.0%; Sales brief: reliably publish-ready 75.0% ROASBench 77 · Overall score 51.8 BulletBench 68 · Ladder Elo 491 (95% range 230–714) Knowledge work 81DeckBench 86 · Deck rating 1,370 (95% range 1,277–1,493) SurveyBench 77 · Analysis rating 1,362 (95% range 1,287–1,462)
No results: Agents and tool use, Writing and preference, Factuality
4
Claude Opus 5.5 Anthropic
80
64 90 89 88 86
$8.00
8 Oct 2026
Spring Prompt benches 64CatalogBench 60 · Reliably publish-ready 55.4%; Sales brief: reliably publish-ready 21.4% ROASBench 71 · Overall score 51.7 Agents and tool use 90Terminal-Bench (Artificial Analysis) 90 · Terminal-Bench 4.0 59.6% Knowledge work 89APEX-Agents 91 · Tasks passed 73.5% GDP.pdf 100 · Professional document tasks 30.6% DeckBench 80 · Deck rating 1,323 (95% range 1,245–1,406) SurveyBench 85 · Analysis rating 1,417 (95% range 1,368–1,478) Writing and preference 88Arena text 95 · Overall 1,504 (95% range 1,495–1,513) UGI Leaderboard 80 · Writing score 70.0 Factuality 86Arena factuality 95 · Overall 1,507 (95% range 1,499–1,515) SimpleQA Verified (Epoch AI) 77 · Correct answers 72.2%
5
GPT-6.1 Sol OpenAI
78
92 81 77 58 66
$4.00
9 Oct 2026
Spring Prompt benches 92CatalogBench 91 · Reliably publish-ready 75.0%; Sales brief: reliably publish-ready 67.9% BulletBench 93 · Lightning 10+1: ladder Elo 584 (95% range 327–806); Bullet 60s: ladder Elo 928 (95% range 676–1,184) Agents and tool use 81Terminal-Bench (Artificial Analysis) 81 · Terminal-Bench 4.0 56.1% Knowledge work 77APEX-Agents 55 · Tasks passed 60.0% DeckBench 89 · Deck rating 1,375 (95% range 1,305–1,463) SurveyBench 88 · Analysis rating 1,429 (95% range 1,364–1,492) Writing and preference 58Arena text 65 · Overall 1,483 (95% range 1,473–1,494) UGI Leaderboard 50 · Writing score 67.7 Factuality 66Arena factuality 40 · Overall 1,463 (95% range 1,454–1,472) SimpleQA Verified (Epoch AI) 92 · Correct answers 73.9%
6
GPT-6.1 Terra OpenAI Provisional · 3 of 5 groups
77
80 77 71 – –
$2.00
27 Oct 2026
Spring Prompt benches 80CatalogBench 82 · Reliably publish-ready 64.3%; Sales brief: reliably publish-ready 67.9% ROASBench 97 · Overall score 58.4 BulletBench 48 · Lightning 10+1: ladder Elo 491 (95% range 0–727); Bullet 60s: ladder Elo 358 (95% range 0–595) Agents and tool use 77Terminal-Bench (Artificial Analysis) 77 · Terminal-Bench 4.0 55.6% Knowledge work 71DeckBench 77 · Deck rating 1,280 (95% range 1,196–1,390) SurveyBench 65 · Analysis rating 1,217 (95% range 1,159–1,294)
No results: Writing and preference, Factuality
7
GPT-6 Astra OpenAI
76
83 79 85 60 65
$20.00
6 Oct 2026
Spring Prompt benches 83CatalogBench 95 · Reliably publish-ready 75.0%; Sales brief: reliably publish-ready 69.6% ROASBench 91 · Overall score 56.6 BulletBench 39 · Ladder Elo 323 (95% range 0–528) Agents and tool use 79Terminal-Bench (Artificial Analysis) 91 · Terminal-Bench 4.0 59.6%; Terminal-Bench 2.1 89.9% τ-bench banking (Artificial Analysis) 45 · Τ-bench banking 43.1% Vending-Bench 2 100 · Money after a year $15,515 Knowledge work 85APEX-Agents 73 · Tasks passed 64.7% DeckBench 97 · Deck rating 1,551 (95% range 1,434–1,730) Writing and preference 60Arena text 50 · Overall 1,477 (95% range 1,470–1,484) UGI Leaderboard 70 · Writing score 68.9 Factuality 65Vectara Hallucination Leaderboard 50 · Hallucination rate 8.7% Arena factuality 45 · Overall 1,466 (95% range 1,460–1,472) SimpleQA Verified (Epoch AI) 100 · Correct answers 75.6%
8
Gemini 4 Argon Google
70
53 85 34 100 100
$4.00
14 Oct 2026
Spring Prompt benches 53CatalogBench 62 · Reliably publish-ready 60.7%; Sales brief: reliably publish-ready 17.9% ROASBench 60 · Overall score 47.0 BulletBench 20 · Ladder Elo 147 (95% range 0–381) Agents and tool use 85Terminal-Bench (Artificial Analysis) 85 · Terminal-Bench 4.0 57.1% Knowledge work 34DeckBench 29 · Deck rating 951 (95% range 883–1,025) SurveyBench 38 · Analysis rating 928 (95% range 863–989) Writing and preference 100Arena text 100 · Overall 1,525 (95% range 1,516–1,534) Factuality 100Arena factuality 100 · Overall 1,511 (95% range 1,503–1,518)
9
Kimi K3.1 Pro Moonshot AI Provisional · 3 of 5 groups
67
61 100 37 – –
$3.81
27 Oct 2026
Spring Prompt benches 61ROASBench 89 · Overall score 55.1 BulletBench 20 · Ladder Elo 147 (95% range 0–381) Agents and tool use 100Terminal-Bench (Artificial Analysis) 100 · Terminal-Bench 4.0 67.2% Knowledge work 37DeckBench 63 · Deck rating 1,223 (95% range 1,095–1,329) SurveyBench 12 · Analysis rating 605 (95% range 462–728)
No results: Writing and preference, Factuality
10=
Muse Wonder Meta Provisional · 3 of 5 groups
66
61 69 76 – –
$8.00
12 Oct 2026
Spring Prompt benches 61CatalogBench 81 · Reliably publish-ready 66.1%; Sales brief: reliably publish-ready 51.8% ROASBench 31 · Overall score 38.9 BulletBench 54 · Ladder Elo 409 (95% range 100–609) Agents and tool use 69Terminal-Bench (Artificial Analysis) 69 · Terminal-Bench 4.0 52.0% Knowledge work 76DeckBench 60 · Deck rating 1,204 (95% range 1,147–1,260) SurveyBench 92 · Analysis rating 1,439 (95% range 1,392–1,508)
No results: Writing and preference, Factuality
10=
GPT-6 Sol OpenAI
66
80 76 63 30 58
$4.00
6 Oct 2026
Spring Prompt benches 80CatalogBench 85 · Reliably publish-ready 66.1%; Sales brief: reliably publish-ready 67.9% ROASBench 86 · Overall score 54.1 BulletBench 59 · Ladder Elo 445 (95% range 215–639) Agents and tool use 76Terminal-Bench (Artificial Analysis) 65 · Terminal-Bench 4.0 43.9% Vending-Bench 2 88 · Money after a year $14,428 Knowledge work 63APEX-Agents 27 · Tasks passed 54.3% GDP.pdf 67 · Professional document tasks 26.4% DeckBench 94 · Deck rating 1,438 (95% range 1,370–1,545) Writing and preference 30Arena text 30 · Overall 1,457 (95% range 1,450–1,464) UGI Leaderboard 30 · Writing score 55.6 Factuality 58Vectara Hallucination Leaderboard 100 · Hallucination rate 6.5% Arena factuality 20 · Overall 1,434 (95% range 1,428–1,440) SimpleQA Verified (Epoch AI) 54 · Correct answers 60.7%
12
Claude Fable 5.1 Anthropic
64
43 74 81 65 80
$20.00
6 Oct 2026
Spring Prompt benches 43CatalogBench 27 · Reliably publish-ready 26.8%; Sales brief: reliably publish-ready 5.4% ROASBench 69 · Overall score 51.5 Agents and tool use 74Terminal-Bench (Artificial Analysis) 87 · Terminal-Bench 4.0 55.1%; Terminal-Bench 2.1 91.4% τ-bench banking (Artificial Analysis) 73 · Τ-bench banking 47.2% Vending-Bench 2 63 · Money after a year $5,422 Knowledge work 81APEX-Agents 82 · Tasks passed 68.6% GDP.pdf 89 · Professional document tasks 29.6% DeckBench 71 · Deck rating 1,256 (95% range 1,191–1,333) Writing and preference 65Arena text 90 · Overall 1,501 (95% range 1,495–1,508) UGI Leaderboard 40 · Writing score 63.9 Factuality 80Arena factuality 90 · Overall 1,500 (95% range 1,494–1,505) SimpleQA Verified (Epoch AI) 69 · Correct answers 70.8%
13
Muse Ember Meta Provisional · 2 of 5 groups
61
61 – 62 – –
$0.20
10 Nov 2026
Spring Prompt benches 61CatalogBench 64 · Reliably publish-ready 53.6%; Sales brief: reliably publish-ready 35.7% ROASBench 74 · Overall score 51.8 BulletBench 34 · Ladder Elo 310 (95% range 0–595) Knowledge work 62DeckBench 51 · Deck rating 1,143 (95% range 1,066–1,226) SurveyBench 73 · Analysis rating 1,345 (95% range 1,284–1,429)
No results: Agents and tool use, Writing and preference, Factuality
14=
Gemini 4 Neon Google Provisional · 2 of 5 groups
60
75 – 25 – –
$6.00
19 Nov 2026
Spring Prompt benches 75CatalogBench 82 · Reliably publish-ready 78.6%; Sales brief: reliably publish-ready 32.1% ROASBench 57 · Overall score 47.0 BulletBench 86 · Ladder Elo 766 (95% range 428–972) Knowledge work 25DeckBench 34 · Deck rating 1,013 (95% range 879–1,113) SurveyBench 15 · Analysis rating 629 (95% range 499–728)
No results: Agents and tool use, Writing and preference, Factuality
14=
Claude Sonnet 5.5 Anthropic
60
43 96 91 40 42
$4.00
6 Oct 2026
Spring Prompt benches 43CatalogBench 36 · Reliably publish-ready 32.1%; Sales brief: reliably publish-ready 10.7% ROASBench 80 · Overall score 52.0 BulletBench 5 · Ladder Elo 33 (95% range 0–270) Agents and tool use 96Terminal-Bench (Artificial Analysis) 96 · Terminal-Bench 4.0 63.6% Knowledge work 91APEX-Agents 100 · Tasks passed 75.5% DeckBench 83 · Deck rating 1,334 (95% range 1,249–1,445) Writing and preference 40Arena text 40 · Overall 1,471 (95% range 1,461–1,481) Factuality 42Arena factuality 60 · Overall 1,470 (95% range 1,461–1,479) SimpleQA Verified (Epoch AI) 23 · Correct answers 46.5%
14=
Gemini 4 Hydrogen Google Provisional · 2 of 5 groups
60
65 – 48 – –
$3.00
16 Nov 2026
Spring Prompt benches 65CatalogBench 67 · Reliably publish-ready 62.5%; Sales brief: reliably publish-ready 23.2% ROASBench 49 · Overall score 45.4 BulletBench 86 · Lightning 10+1: ladder Elo 557 (95% range 370–727); Bullet 60s: ladder Elo 889 (95% range 593–1,229) Knowledge work 48DeckBench 43 · Deck rating 1,079 (95% range 983–1,176) SurveyBench 54 · Analysis rating 1,096 (95% range 997–1,190)
No results: Agents and tool use, Writing and preference, Factuality
17
Muse Spark 1.3 Meta
58
30 78 49 80 85
$2.00
6 Oct 2026
Spring Prompt benches 30CatalogBench 45 · Reliably publish-ready 48.2%; Sales brief: reliably publish-ready 10.7% ROASBench 23 · Overall score 34.1 BulletBench 4 · Ladder Elo 33 (95% range 0–270) Agents and tool use 78Terminal-Bench (Artificial Analysis) 57 · Terminal-Bench 4.0 33.3%; Terminal-Bench 2.1 85.4% τ-bench banking (Artificial Analysis) 100 · Τ-bench banking 50.5% Knowledge work 49APEX-Agents 45 · Tasks passed 57.8% GDP.pdf 78 · Professional document tasks 27.6% DeckBench 23 · Deck rating 807 (95% range 748–854) Writing and preference 80Arena text 80 · Overall 1,494 (95% range 1,488–1,500) Factuality 85Arena factuality 85 · Overall 1,480 (95% range 1,475–1,485)
18
Gemini 3.8 Flash Google
57
47 47 43 93 71
$3.00
10 Nov 2026
Spring Prompt benches 47CatalogBench 30 · Reliably publish-ready 35.7%; Sales brief: reliably publish-ready 3.6% ROASBench 66 · Overall score 51.4 BulletBench 60 · Lightning 10+1: ladder Elo 197 (95% range 0–439); Bullet 60s: ladder Elo 1,068 (95% range 888–1,241) Agents and tool use 47Terminal-Bench (Artificial Analysis) 50 · Terminal-Bench 4.0 19.7%; Terminal-Bench 2.1 87.6% τ-bench banking (Artificial Analysis) 55 · Τ-bench banking 45.8% Vending-Bench 2 38 · Money after a year $5,094 Knowledge work 43APEX-Agents 64 · Tasks passed 64.3% GDP.pdf 56 · Professional document tasks 23.4% DeckBench 20 · Deck rating 771 (95% range 706–825) SurveyBench 35 · Analysis rating 845 (95% range 761–922) Writing and preference 93Arena text 85 · Overall 1,495 (95% range 1,490–1,500) UGI Leaderboard 100 · Writing score 78.6 Factuality 71Arena factuality 80 · Overall 1,477 (95% range 1,472–1,481) SimpleQA Verified (Epoch AI) 62 · Correct answers 69.7%
19
Qwen4-Max Alibaba Provisional · 2 of 5 groups
53
37 – 90 – –
$2.25
16 Nov 2026
Spring Prompt benches 37CatalogBench 35 · Reliably publish-ready 39.3%; Sales brief: reliably publish-ready 8.9% ROASBench 51 · Overall score 45.8 BulletBench 20 · Ladder Elo 147 (95% range 0–381) Knowledge work 90DeckBench 100 · Deck rating 1,563 (95% range 1,478–1,692) SurveyBench 81 · Analysis rating 1,381 (95% range 1,318–1,464)
No results: Agents and tool use, Writing and preference, Factuality
20
Gemini 4 Carbon Google Provisional · 2 of 5 groups
50
54 – 39 – –
$8.00
11 Nov 2026
Spring Prompt benches 54CatalogBench 52 · Reliably publish-ready 53.6%; Sales brief: reliably publish-ready 14.3% ROASBench 63 · Overall score 51.2 BulletBench 46 · Ladder Elo 358 (95% range 0–619) Knowledge work 39DeckBench 31 · Deck rating 992 (95% range 851–1,114) SurveyBench 46 · Analysis rating 1,054 (95% range 961–1,162)
No results: Agents and tool use, Writing and preference, Factuality
21
DeepSeek-V4.1-Flash DeepSeek
49
51 42 – 45 55
$0.52
30 Sep 2026
Spring Prompt benches 51CatalogBench 51 · Reliably publish-ready 41.1%; Sales brief: reliably publish-ready 23.2% Agents and tool use 42Terminal-Bench (Artificial Analysis) 42 · Terminal-Bench 4.0 26.8% Writing and preference 45Arena text 45 · Overall 1,474 (95% range 1,468–1,481) Factuality 55Arena factuality 55 · Overall 1,467 (95% range 1,461–1,473)
No results: Knowledge work
22
Kimi K3 Moonshot AI
46
30 50 33 75 61
$6.00
6 Oct 2026
Spring Prompt benches 30CatalogBench 31 · Reliably publish-ready 41.1%; Sales brief: reliably publish-ready 1.8% ROASBench 46 · Overall score 44.1 BulletBench 4 · Ladder Elo 33 (95% range 0–270) Agents and tool use 50Terminal-Bench (Artificial Analysis) 36 · Terminal-Bench 4.0 12.6%; Terminal-Bench 2.1 85.0% τ-bench banking (Artificial Analysis) 64 · Τ-bench banking 46.0% Vending-Bench 2 50 · Money after a year $5,165 Knowledge work 33APEX-Agents 18 · Tasks passed 50.6% GDP.pdf 22 · Professional document tasks 19.0% DeckBench 57 · Deck rating 1,194 (95% range 1,148–1,277) Writing and preference 75Arena text 75 · Overall 1,488 (95% range 1,484–1,493) Factuality 61Arena factuality 75 · Overall 1,472 (95% range 1,469–1,476) SimpleQA Verified (Epoch AI) 46 · Correct answers 50.6%
23=
Gemini 3.1 Pro Preview Google
44
52 21 13 80 52
$4.50
8 Oct 2026
Spring Prompt benches 52CatalogBench 49 · Reliably publish-ready 44.6%; Sales brief: reliably publish-ready 16.1% ROASBench 43 · Overall score 43.5 BulletBench 75 · Ladder Elo 529 (95% range 329–698) Agents and tool use 21Terminal-Bench (Artificial Analysis) 24 · Terminal-Bench 4.0 4.0%; Terminal-Bench 2.1 73.8% τ-bench banking (Artificial Analysis) 27 · Τ-bench banking 21.4% Vending-Bench 2 13 · Money after a year $911.21 Knowledge work 13APEX-Agents 0 · Tasks passed 35.3% GDP.pdf 11 · Professional document tasks 17.0% DeckBench 14 · Deck rating 716 (95% range 644–787) SurveyBench 27 · Analysis rating 744 (95% range 687–795) Writing and preference 80Arena text 70 · Overall 1,487 (95% range 1,484–1,490) UGI Leaderboard 90 · Writing score 72.2 Factuality 52Vectara Hallucination Leaderboard 0 · Hallucination rate 10.4% Arena factuality 70 · Overall 1,472 (95% range 1,470–1,474) SimpleQA Verified (Epoch AI) 85 · Correct answers 73.5%
23=
Qwen3.8-Max (0902) Alibaba
44
17 76 41 60 52
$3.00
6 Oct 2026
Spring Prompt benches 17CatalogBench 18 · Reliably publish-ready 28.6%; Sales brief: reliably publish-ready 0.0% ROASBench 14 · Overall score 25.4 Agents and tool use 76Terminal-Bench (Artificial Analysis) 70 · Terminal-Bench 4.0 38.9%; Terminal-Bench 2.1 88.8% τ-bench banking (Artificial Analysis) 82 · Τ-bench banking 47.8% Knowledge work 41GDP.pdf 44 · Professional document tasks 23.2% DeckBench 37 · Deck rating 1,025 (95% range 954–1,102) Writing and preference 60Arena text 60 · Overall 1,482 (95% range 1,476–1,487) Factuality 52Arena factuality 65 · Overall 1,472 (95% range 1,467–1,476) SimpleQA Verified (Epoch AI) 38 · Correct answers 47.3%
25
MiniMax M3.1 Flash MiniMax Provisional · 3 of 5 groups
40
44 37 37 – –
$0.26
4 Nov 2026
Spring Prompt benches 44CatalogBench 63 · Reliably publish-ready 50.0%; Sales brief: reliably publish-ready 37.5% ROASBench 34 · Overall score 40.3 BulletBench 11 · Ladder Elo 115 (95% range 0–485) Agents and tool use 37Terminal-Bench (Artificial Analysis) 37 · Terminal-Bench 4.0 25.8% Knowledge work 37DeckBench 74 · Deck rating 1,266 (95% range 1,177–1,367) SurveyBench 0 · Analysis rating 428 (95% range 220–567)
No results: Writing and preference, Factuality
26
Grok 4.7 xAI
39
52 37 66 18 5
$3.00
6 Oct 2026
Spring Prompt benches 52CatalogBench 68 · Reliably publish-ready 57.1%; Sales brief: reliably publish-ready 33.9% ROASBench 26 · Overall score 36.6 Agents and tool use 37Terminal-Bench (Artificial Analysis) 37 · Terminal-Bench 4.0 25.8% Knowledge work 66DeckBench 66 · Deck rating 1,242 (95% range 1,173–1,326) Writing and preference 18Arena text 15 · Overall 1,442 (95% range 1,435–1,450) UGI Leaderboard 20 · Writing score 54.1 Factuality 5Arena factuality 5 · Overall 1,428 (95% range 1,421–1,435)
27=
Claude Haiku 5.5 Anthropic Provisional · 3 of 5 groups
38
25 46 59 – –
$0.20
8 Oct 2026
Spring Prompt benches 25CatalogBench 32 · Reliably publish-ready 26.8%; Sales brief: reliably publish-ready 10.7% ROASBench 6 · Overall score 24.1 BulletBench 34 · Ladder Elo 310 (95% range 0–575) Agents and tool use 46Terminal-Bench (Artificial Analysis) 46 · Terminal-Bench 4.0 32.8% Knowledge work 59DeckBench 49 · Deck rating 1,120 (95% range 1,050–1,188) SurveyBench 69 · Analysis rating 1,233 (95% range 1,194–1,291)
No results: Writing and preference, Factuality
27=
GPT-6 Luna OpenAI
38
67 17 40 20 13
$0.20
10 Nov 2026
Spring Prompt benches 67CatalogBench 72 · Reliably publish-ready 55.4%; Sales brief: reliably publish-ready 51.8% ROASBench 83 · Overall score 53.4 BulletBench 29 · Ladder Elo 249 (95% range 0–453) Agents and tool use 17Terminal-Bench (Artificial Analysis) 17 · Terminal-Bench 4.0 12.6% Knowledge work 40APEX-Agents 9 · Tasks passed 44.3% GDP.pdf 33 · Professional document tasks 23.0% DeckBench 69 · Deck rating 1,252 (95% range 1,194–1,333) SurveyBench 50 · Analysis rating 1,072 (95% range 975–1,156) Writing and preference 20Arena text 20 · Overall 1,443 (95% range 1,436–1,449) Factuality 13Arena factuality 10 · Overall 1,430 (95% range 1,424–1,436) SimpleQA Verified (Epoch AI) 15 · Correct answers 41.4%
29
DeepSeek-V4.1-Flash-1017 DeepSeek Provisional · 3 of 5 groups
37
22 54 49 – –
$0.52
19 Oct 2026
Spring Prompt benches 22CatalogBench 14 · Reliably publish-ready 21.4%; Sales brief: reliably publish-ready 0.0% ROASBench 37 · Overall score 41.4 BulletBench 20 · Ladder Elo 147 (95% range 0–443) Agents and tool use 54Terminal-Bench (Artificial Analysis) 54 · Terminal-Bench 4.0 33.8% Knowledge work 49DeckBench 40 · Deck rating 1,046 (95% range 944–1,120) SurveyBench 58 · Analysis rating 1,119 (95% range 1,045–1,196)
No results: Writing and preference, Factuality
30
GLM-5.3 Z.ai
35
7 73 36 55 29
$2.15
1 Oct 2026
Spring Prompt benches 7ROASBench 9 · Overall score 24.8 BulletBench 4 · Ladder Elo 33 (95% range 0–270) Agents and tool use 73Terminal-Bench (Artificial Analysis) 53 · Terminal-Bench 4.0 41.9%; Terminal-Bench 2.1 83.9% τ-bench banking (Artificial Analysis) 91 · Τ-bench banking 50.3% Vending-Bench 2 75 · Money after a year $8,164 Knowledge work 36APEX-Agents 36 · Tasks passed 56.6% Writing and preference 55Arena text 55 · Overall 1,478 (95% range 1,473–1,484) Factuality 29Arena factuality 50 · Overall 1,467 (95% range 1,462–1,472) SimpleQA Verified (Epoch AI) 8 · Correct answers 41.0%
31
Mercury 3 Inception Provisional · 2 of 5 groups
33
38 – 20 – –
–
19 Oct 2026
Spring Prompt benches 38ROASBench 17 · Overall score 26.1 BulletBench 69 · Lightning 10+1: ladder Elo 555 (95% range 263–727); Bullet 60s: ladder Elo 491 (95% range 299–659) Knowledge work 20DeckBench 17 · Deck rating 726 (95% range 632–814) SurveyBench 23 · Analysis rating 735 (95% range 631–817)
No results: Agents and tool use, Writing and preference, Factuality
32=
GLM-5.3 FlashX Legal Z.ai Provisional · 2 of 5 groups
29
27 – 34 – –
$0.59
4 Nov 2026
Spring Prompt benches 27CatalogBench 4 · Reliably publish-ready 0.0%; Sales brief: reliably publish-ready 0.0% ROASBench 54 · Overall score 46.2 BulletBench 43 · Lightning 10+1: ladder Elo 310 (95% range 0–641); Bullet 60s: ladder Elo 358 (95% range 0–600) Knowledge work 34DeckBench 26 · Deck rating 809 (95% range 678–919) SurveyBench 42 · Analysis rating 937 (95% range 830–1,019)
No results: Agents and tool use, Writing and preference, Factuality
32=
Particle-1 Particle AI Provisional · 2 of 5 groups
29
18 – 54 – –
$0.16
4 Nov 2026
Spring Prompt benches 18CatalogBench 4 · Reliably publish-ready 0.0%; Sales brief: reliably publish-ready 0.0% ROASBench 40 · Overall score 43.3 Knowledge work 54DeckBench 46 · Deck rating 1,086 (95% range 983–1,185) SurveyBench 62 · Analysis rating 1,133 (95% range 1,072–1,222)
No results: Agents and tool use, Writing and preference, Factuality
34
DeepSeek-V4-Pro (0423) DeepSeek
27
20 29 0 48 47
$1.77
6 Oct 2026
Spring Prompt benches 20ROASBench 20 · Overall score 28.3 Agents and tool use 29Terminal-Bench (Artificial Analysis) 25 · Terminal-Bench 4.0 14.6%; Terminal-Bench 2.1 64.8% τ-bench banking (Artificial Analysis) 36 · Τ-bench banking 30.1% Vending-Bench 2 25 · Money after a year $3,285 Knowledge work 0DeckBench 0 · Deck rating 530 (95% range 391–611) Writing and preference 48Arena text 35 · Overall 1,458 (95% range 1,454–1,462) UGI Leaderboard 60 · Writing score 68.4 Factuality 47Vectara Hallucination Leaderboard 75 · Hallucination rate 8.6% Arena factuality 35 · Overall 1,455 (95% range 1,452–1,458) SimpleQA Verified (Epoch AI) 31 · Correct answers 47.0%
35=
Gemini 3.5 Flash-Lite Google
24
33 16 9 25 30
$0.85
8 Oct 2026
Spring Prompt benches 33CatalogBench 21 · Reliably publish-ready 17.9%; Sales brief: reliably publish-ready 3.6% ROASBench 11 · Overall score 25.4 BulletBench 95 · Lightning 10+1: ladder Elo 892 (95% range 641–1,081); Bullet 60s: ladder Elo 780 (95% range 586–956) Agents and tool use 16Terminal-Bench (Artificial Analysis) 13 · Terminal-Bench 4.0 1.0%; Terminal-Bench 2.1 53.6% τ-bench banking (Artificial Analysis) 18 · Τ-bench banking 17.5% Knowledge work 9GDP.pdf 0 · Professional document tasks 10.0% DeckBench 9 · Deck rating 605 (95% range 494–695) SurveyBench 19 · Analysis rating 637 (95% range 580–689) Writing and preference 25Arena text 25 · Overall 1,455 (95% range 1,450–1,459) Factuality 30Arena factuality 30 · Overall 1,447 (95% range 1,444–1,451)
35=
Mistral Large 4 Mistral AI Provisional · 3 of 5 groups
24
22 31 18 – –
$2.06
8 Oct 2026
Spring Prompt benches 22CatalogBench 15 · Reliably publish-ready 10.7%; Sales brief: reliably publish-ready 1.8% ROASBench 29 · Overall score 37.0 BulletBench 30 · Ladder Elo 249 (95% range 0–486) Agents and tool use 31Terminal-Bench (Artificial Analysis) 31 · Terminal-Bench 4.0 24.7% Knowledge work 18DeckBench 6 · Deck rating 595 (95% range 472–693) SurveyBench 31 · Analysis rating 770 (95% range 716–811)
No results: Writing and preference, Factuality
37
Mistral Medium 3.5 Mistral AI
15
28 7 5 3 15
$3.00
8 Oct 2026
Spring Prompt benches 28CatalogBench 28 · Reliably publish-ready 17.9%; Sales brief: reliably publish-ready 10.7% ROASBench 0 · Overall score 15.2 BulletBench 71 · Lightning 10+1: ladder Elo 529 (95% range 361–669); Bullet 60s: ladder Elo 572 (95% range 428–708) Agents and tool use 7Terminal-Bench (Artificial Analysis) 6 · Terminal-Bench 4.0 0.0%; Terminal-Bench 2.1 50.6% τ-bench banking (Artificial Analysis) 9 · Τ-bench banking 15.1% Knowledge work 5DeckBench 3 · Deck rating 585 (95% range 473–662) SurveyBench 8 · Analysis rating 578 (95% range 495–641) Writing and preference 3Arena text 5 · Overall 1,427 (95% range 1,421–1,433) UGI Leaderboard 0 · Writing score 45.5 Factuality 15Arena factuality 15 · Overall 1,431 (95% range 1,425–1,436)
38
GLM-5V-Turbo Z.ai Provisional · 3 of 5 groups
12
7 – – 10 25
$1.90
30 Sep 2026
Spring Prompt benches 7CatalogBench 7 · Reliably publish-ready 1.8%; Sales brief: reliably publish-ready 0.0% Writing and preference 10Arena text 10 · Overall 1,433 (95% range 1,427–1,440) Factuality 25Arena factuality 25 · Overall 1,440 (95% range 1,435–1,446)
No results: Agents and tool use, Knowledge work
39
Claude Haiku 4.5 Anthropic
7
11 0 8 5 8
$2.00
8 Oct 2026
Spring Prompt benches 11CatalogBench 7 · Reliably publish-ready 1.8%; Sales brief: reliably publish-ready 0.0% ROASBench 3 · Overall score 17.5 BulletBench 32 · Lightning 10+1: ladder Elo 33 (95% range 0–248); Bullet 60s: ladder Elo 445 (95% range 135–626) Agents and tool use 0Terminal-Bench (Artificial Analysis) 1 · Terminal-Bench 4.0 0.0%; Terminal-Bench 2.1 44.2% τ-bench banking (Artificial Analysis) 0 · Τ-bench banking 9.3% Vending-Bench 2 0 · Money after a year $458.89 Knowledge work 8DeckBench 11 · Deck rating 646 (95% range 541–738) SurveyBench 4 · Analysis rating 561 (95% range 482–618) Writing and preference 5Arena text 0 · Overall 1,414 (95% range 1,411–1,416) UGI Leaderboard 10 · Writing score 52.2 Factuality 8Vectara Hallucination Leaderboard 25 · Hallucination rate 9.8% Arena factuality 0 · Overall 1,418 (95% range 1,416–1,420) SimpleQA Verified (Epoch AI) 0 · Correct answers 13.2%