Math · Latency

Latency on Math

2,163 questions, asked in English. How to read this page · Method

Latency

Milliseconds from request to reply per question (log scale): the median (family colour) and the 90th percentile (grey, nearest rank); beside them questions per minute, wall clock from the run's first to last answer, and requests in flight on average over its active time · lower latency is better

  1. GLiDE 0.4B

    GLiDE 0.4B · Fastino · decision model

    Parameters: 0.40B dense

    Latency: median 11 ms · p90 12 ms

    Logic 23.2%

    Its run record (Method) →

    11 ms12 ms4,632.9≈ 0.8
  2. GLiClass v4 Edge

    GLiClass v4 Edge · decision model

    Parameters: 0.4B dense

    Base model: DeBERTa-v3-large

    Latency: median 12 ms · p90 13 ms

    Logic 23.6%

    Its run record (Method) →

    12 ms13 ms4,184.5≈ 0.9
  3. OneJev1.6-0.8B

    OneJev1.6-0.8B · OmniJev · decision model

    Parameters: 0.8B dense

    Latency: median 15 ms · p90 17 ms

    Logic 27.0%

    Its run record (Method) →

    15 ms17 ms3,243.0≈ 0.8
  4. d1-2B

    d1-2B · Liquid AI · decision model

    Parameters: 1.92B dense

    Base model: LFM3-VL-2B

    Latency: median 16 ms · p90 21 ms

    Logic 30.7%

    Its run record (Method) →

    16 ms21 ms2,948.2≈ 0.8
  5. GLiDE 0.8B

    GLiDE 0.8B · Fastino · decision model

    Parameters: 0.75B dense

    Latency: median 20 ms · p90 21 ms

    Logic 29.2%

    Its run record (Method) →

    20 ms21 ms2,760.0≈ 0.9
  6. StartLux-Decision-v1.1-0.8B-Instant

    StartLux-Decision-v1.1-0.8B-Instant · StartLux · decision model

    Parameters: 0.8B dense

    Base model: Qwen3.5-0.8B

    Latency: median 24 ms · p90 25 ms

    Logic 24.7%

    Its run record (Method) →

    24 ms25 ms2,316.4≈ 0.9
  7. KnowLine-0.8B-Gen1

    KnowLine-0.8B-Gen1 · PelaAI · decision model

    Parameters: 0.8B dense

    Base model: Qwen3.5-0.8B

    Latency: median 25 ms · p90 28 ms

    Logic 26.7%

    Its run record (Method) →

    25 ms28 ms2,092.3≈ 0.9
  8. GLiDE 2B

    GLiDE 2B · Fastino · decision model

    Parameters: 1.88B dense

    Latency: median 28 ms · p90 30 ms

    Logic 32.3%

    Its run record (Method) →

    28 ms30 ms1,995.7≈ 0.9
  9. Perplexity Decider Flash v1.3

    Perplexity Decider Flash v1.3 · Perplexity · decision model

    Parameters: 4B dense

    Latency: median 28 ms · p90 31 ms

    Logic 44.7%

    Its run record (Method) →

    28 ms31 ms1,853.1≈ 0.9
  10. Chinese Decision Pointer 4B

    Chinese Decision Pointer 4B · quincyqiang · decision model

    Parameters: 4B dense

    Base model: Qwen3.5-4B

    Latency: median 28 ms · p90 32 ms

    Logic 40.4%

    Its run record (Method) →

    28 ms32 ms1,827.0≈ 0.9
  11. vera-medium-omni

    vera-medium-omni · scar-ai · decision model

    Parameters: 4.2B dense

    Base model: Qwen3.5-4B

    Latency: median 31 ms · p90 35 ms

    Logic 44.1%

    Its run record (Method) →

    31 ms35 ms1,777.0≈ 0.9
  12. Strands Decider 2B

    Strands Decider 2B · Strands Agents · decision model

    Parameters: 2B dense

    Latency: median 35 ms · p90 41 ms

    Logic 22.4%

    Its run record (Method) →

    35 ms41 ms1,582.0≈ 0.9
  13. StartLux-Decision-v1.1-2B-Instant

    StartLux-Decision-v1.1-2B-Instant · StartLux · decision model

    Parameters: 2B dense

    Latency: median 36 ms · p90 37 ms

    Logic 30.8%

    Its run record (Method) →

    36 ms37 ms1,562.9≈ 0.9
  14. Adjudex 9B v5

    Adjudex 9B v5 · newsbang · decision model

    Parameters: 9B dense

    Base model: Qwen3.5-9B

    Latency: median 36 ms · p90 44 ms

    Logic 46.4%

    Its run record (Method) →

    36 ms44 ms1,562.9≈ 0.9
  15. Jeffan-9B-1211

    Jeffan-9B-1211 · tattrongvu · decision model

    Parameters: 9B dense

    Base model: Qwen3.5-9B

    Latency: median 37 ms · p90 51 ms

    Logic 48.2%

    Its run record (Method) →

    37 ms51 ms1,441.3≈ 0.9
  16. KnowLine-9B-Gen1

    KnowLine-9B-Gen1 · PelaAI · decision model

    Parameters: 9B dense

    Base model: Qwen3.5-9B

    Latency: median 37 ms · p90 43 ms

    Logic 44.3%

    Its run record (Method) →

    37 ms43 ms1,562.9≈ 1.0
  17. Cygnet

    Cygnet · Blockbrain Labs · decision model

    Parameters: 11.95B dense

    Base model: Gemma 4 12B IT

    Latency: median 37 ms · p90 41 ms

    Logic 39.3%

    Its run record (Method) →

    37 ms41 ms1,508.4≈ 0.9
  18. Perplexity Decider Flash v1.2

    Perplexity Decider Flash v1.2 · Perplexity · decision model

    Parameters: 4B dense

    Latency: median 40 ms · p90 47 ms

    Logic 43.1%

    Its run record (Method) →

    40 ms47 ms1,380.0≈ 0.9
  19. KnowLine-9B-Gen2

    KnowLine-9B-Gen2 · PelaAI · decision model

    Parameters: 9B dense

    Base model: Qwen3.5-9B

    Latency: median 41 ms · p90 46 ms

    Logic 43.9%

    Its run record (Method) →

    41 ms46 ms1,337.3≈ 0.9
  20. Dexy-Decision-9B-V2

    Dexy-Decision-9B-V2 · Patrick-18 · decision model

    Parameters: 9B dense

    Base model: vllm-sr/d3-flash

    Latency: median 45 ms · p90 50 ms

    Logic 45.3%

    Its run record (Method) →

    45 ms50 ms1,297.2≈ 0.9
  21. Verdin-4B-v1.1

    Verdin-4B-v1.1 · victorbennett · decision model

    Parameters: 4B dense

    Base model: Qwen3.5-4B

    Latency: median 47 ms · p90 52 ms

    Logic 39.7%

    Its run record (Method) →

    47 ms52 ms1,235.4≈ 1.0
  22. KnowLine-12B-Gen1

    KnowLine-12B-Gen1 · PelaAI · decision model

    Parameters: 12B dense

    Base model: Gemma 4 12B

    Latency: median 49 ms · p90 55 ms

    Logic 50.6%

    Its run record (Method) →

    49 ms55 ms1,148.0≈ 0.9
  23. KnowLine-2B-Gen1

    KnowLine-2B-Gen1 · PelaAI · decision model

    Parameters: 2B dense

    Base model: Qwen3.5-2B

    Latency: median 50 ms · p90 54 ms

    Logic 31.7%

    Its run record (Method) →

    50 ms54 ms1,108.7≈ 0.9
  24. Winnow-12B

    Winnow-12B · EldanRing · decision model

    Parameters: 12B dense

    Latency: median 52 ms · p90 61 ms

    Logic 39.7%

    Its run record (Method) →

    52 ms61 ms1,081.0≈ 1.0
  25. d3

    d3 · vLLM Semantic Router · decision model

    Parameters: 26.09B dense

    Latency: median 55 ms · p90 62 ms

    Logic 54.5%

    Its run record (Method) →

    55 ms62 ms1,021.4≈ 1.0
  26. Valen-Decision-1.5-9B

    Valen-Decision-1.5-9B · Valen-Team · decision model

    Parameters: 9B dense

    Base model: Qwen3.5-9B

    Latency: median 58 ms · p90 64 ms

    Logic 44.4%

    Its run record (Method) →

    58 ms64 ms1,013.4≈ 1.0
  27. StartLux-Decision-v1.1-4B-Instant

    StartLux-Decision-v1.1-4B-Instant · StartLux · decision model

    Parameters: 4B dense

    Latency: median 67 ms · p90 69 ms

    Logic 45.9%

    Its run record (Method) →

    67 ms69 ms864.8≈ 1.0
  28. KnowLine 4B Gen5.5

    KnowLine 4B Gen5.5 · PelaAI · decision model

    Parameters: 4B dense

    Latency: median 74 ms · p90 84 ms

    Logic 42.3%

    Its run record (Method) →

    74 ms84 ms776.8≈ 1.0
  29. Drex Max

    Drex Max · Nace.AI · decision model

    Parameters: 27B dense

    Latency: median 76 ms · p90 89 ms

    Logic 53.1%

    Its run record (Method) →

    76 ms89 ms758.6≈ 1.0
  30. GLiDE v1.2

    GLiDE v1.2 · Fastino · decision model

    Parameters: 27B dense

    Latency: median 79 ms · p90 84 ms

    Logic 55.2%

    Its run record (Method) →

    79 ms84 ms716.7≈ 1.0
  31. OneJev-27B

    OneJev-27B · OmniJev · decision model

    Parameters: 27B dense

    Latency: median 79 ms · p90 106 ms

    Logic 40.6%

    Its run record (Method) →

    79 ms106 ms682.7≈ 1.0
  32. Elephant-1

    Elephant-1 · Base Labs · decision model

    Parameters: 9B dense

    Latency: median 80 ms · p90 92 ms

    Logic 46.4%

    Its run record (Method) →

    80 ms92 ms712.7≈ 1.0
  33. Perplexity Decider 27B v1.3 NVIDIA RTX PRO 6000 Blackwell

    Perplexity Decider 27B v1.3 NVIDIA RTX PRO 6000 Blackwell · Perplexity · decision model

    Parameters: 27B dense

    Latency: median 91 ms · p90 103 ms

    Logic 54.8%

    Its run record (Method) →

    91 ms103 ms629.7≈ 1.0
  34. Clef-Flash

    Clef-Flash · Cloudflare · decision model

    Parameters: 9B dense

    Latency: median 99 ms · p90 127 ms

    Logic 38.9%

    Its run record (Method) →

    99 ms127 ms576.5≈ 1.0
  35. Metask-Jev-4B

    Metask-Jev-4B · Metask Lab · decision model

    Parameters: 4.54B dense

    Latency: median 113 ms · p90 144 ms

    Logic 32.1%

    Its run record (Method) →

    113 ms144 ms498.9≈ 1.0
  36. KnowLine-31B-Gen2

    KnowLine-31B-Gen2 · PelaAI · decision model

    Parameters: 31B dense

    Base model: Gemma 4 31B

    Latency: median 114 ms · p90 120 ms

    Logic 58.9%

    Its run record (Method) →

    114 ms120 ms498.9≈ 0.9
  37. Copernicus-31B

    Copernicus-31B · Poindexter Labs · decision model

    Parameters: 31B dense

    Latency: median 118 ms · p90 142 ms

    Logic 58.7%

    Its run record (Method) →

    118 ms142 ms491.4≈ 1.0
  38. d3-ultra

    d3-ultra · vLLM Semantic Router · decision model

    Parameters: 26.09B dense

    Latency: median 118 ms · p90 148 ms

    Logic 54.6%

    Its run record (Method) →

    118 ms148 ms13.6≈ 1.0
  39. Mammoth-1

    Mammoth-1 · Base Labs · decision model

    Parameters: 27B dense

    Latency: median 140 ms · p90 173 ms

    Logic 54.5%

    Its run record (Method) →

    140 ms173 ms399.1≈ 1.0
  40. StartLux-Decision-v1.1-9B-Instant

    StartLux-Decision-v1.1-9B-Instant · StartLux · decision model

    Parameters: 9B dense

    Latency: median 140 ms · p90 146 ms

    Logic 46.6%

    Its run record (Method) →

    140 ms146 ms425.3≈ 1.0
  41. Cerno 9B v1.2

    Cerno 9B v1.2 · Apertix AI · decision model

    Parameters: 9B dense

    Latency: median 153 ms · p90 175 ms

    Logic 43.3%

    Its run record (Method) →

    153 ms175 ms372.8≈ 1.0
  42. Karpo Decision 31B

    Karpo Decision 31B · MachinePulse · decision model

    Parameters: 31B dense

    Latency: median 155 ms · p90 186 ms

    Logic 60.7%

    Its run record (Method) →

    155 ms186 ms368.5≈ 1.0
  43. Jev 1.14

    Jev 1.14 · TypeSafe · decision model

    Parameters: not disclosed by maker

    Latency: median 163 ms · p90 185 ms

    Logic 48.2%

    Its run record (Method) →

    163 ms185 ms323.5≈ 0.9
  44. Jev 1.13

    Jev 1.13 · TypeSafe · decision model

    Parameters: not disclosed by maker

    Latency: median 166 ms · p90 206 ms

    Logic 46.3%

    Its run record (Method) →

    166 ms206 ms333.5≈ 1.0
  45. d1.1

    d1.1 · Liquid AI · decision model

    Parameters: not disclosed by maker

    Latency: median 172 ms · p90 290 ms

    Logic 46.0%

    Its run record (Method) →

    172 ms290 ms285.1≈ 1.0
  46. GLiDE 9B

    GLiDE 9B · Fastino · decision model

    Parameters: 7.94B dense

    Latency: median 174 ms · p90 182 ms

    Logic 47.8%

    Its run record (Method) →

    174 ms182 ms337.8≈ 1.0
  47. d1-ultra

    d1-ultra · Liquid AI · decision model

    Parameters: not disclosed by maker (built on a larger open-weight model, not named with a count)

    Latency: median 179 ms · p90 426 ms

    Logic 54.9%

    Its run record (Method) →

    179 ms426 ms250.9≈ 1.0
  48. Clef

    Clef · Cloudflare · decision model

    Parameters: 27B dense

    Latency: median 181 ms · p90 210 ms

    Logic 40.8%

    Its run record (Method) →

    181 ms210 ms315.6≈ 1.0
  49. d1

    d1 · Liquid AI · decision model

    Parameters: not disclosed by maker

    Latency: median 184 ms · p90 297 ms

    Logic 40.8%

    Its run record (Method) →

    184 ms297 ms278.4≈ 1.0
  50. REAX 1.3

    REAX 1.3 · REAX · decision model

    Parameters: 57.36B dense

    Latency: median 189 ms · p90 232 ms

    Logic 59.3%

    Its run record (Method) →

    189 ms232 ms292.2≈ 1.0
  51. Logits Sigma-1 Duo

    Logits Sigma-1 Duo · Logits · decision model

    Parameters: 57.36B dense

    Latency: median 196 ms · p90 221 ms

    Logic 57.8%

    Its run record (Method) →

    196 ms221 ms294.8≈ 1.0
  52. d1.2

    d1.2 · Liquid AI · decision model

    Parameters: not disclosed by maker

    Latency: median 204 ms · p90 386 ms

    Logic 49.2%

    Its run record (Method) →

    204 ms386 ms236.7≈ 1.0
  53. d1.2-ultra

    d1.2-ultra · Liquid AI · decision model

    Parameters: not disclosed by maker

    Latency: median 215 ms · p90 347 ms

    Logic 51.6%

    Its run record (Method) →

    215 ms347 ms229.6≈ 1.0
  54. Gemini Decision 1 20261111

    Gemini Decision 1 20261111 · Google · decision model

    Parameters: not disclosed by maker (closed weights)

    Latency: median 226 ms · p90 351 ms

    Logic 50.0%

    Its run record (Method) →

    226 ms351 ms98.9≈ 0.5
  55. Perplexity Decider 27B v1.2 API

    Perplexity Decider 27B v1.2 API · Perplexity · decision model

    Parameters: 27B dense

    Latency: median 226 ms · p90 368 ms

    Logic 53.3%

    Its run record (Method) →

    226 ms368 ms217.3≈ 1.0
  56. Open-Jev-27B-v1.1

    Open-Jev-27B-v1.1 · Zefan Cai · decision model

    Parameters: 27B dense

    Base model: Qwen3.8-27B + adapter

    Latency: median 233 ms · p90 272 ms

    Logic 42.6%

    Its run record (Method) →

    233 ms272 ms245.2≈ 1.0
  57. GLiDE API

    GLiDE API · Fastino · decision model

    Parameters: not disclosed by maker

    Latency: median 264 ms · p90 279 ms

    Logic 54.5%

    Its run record (Method) →

    264 ms279 ms882.4≈ 4.0
  58. Turion Jev 31B

    Turion Jev 31B · Turion.AI · decision model

    Parameters: 31B dense

    Latency: median 278 ms · p90 329 ms

    Logic 52.3%

    Its run record (Method) →

    278 ms329 ms203.0≈ 1.0
  59. Mercury Decide

    Mercury Decide · Inception · decision model

    Parameters: not disclosed by maker

    Latency: median 288 ms · p90 342 ms

    Logic 45.1%

    Its run record (Method) →

    288 ms342 ms196.2≈ 1.0
  60. Microsoft-Decision-1

    Microsoft-Decision-1 · Microsoft · decision model

    Parameters: not disclosed by maker

    Base model: Qwen3.5-9B

    Latency: median 310 ms · p90 359 ms

    Logic 44.4%

    Its run record (Method) →

    310 ms359 ms184.8≈ 1.0
  61. Drex Max v1.1

    Drex Max v1.1 · Nace.AI · decision model

    Parameters: not disclosed by maker

    Latency: median 326 ms · p90 473 ms

    Logic 54.8%

    Its run record (Method) →

    326 ms473 ms118.0≈ 1.0
  62. Drex 2.0

    Drex 2.0 · Nace.AI · decision model

    Parameters: 35B MoE · 3B active

    Latency: median 334 ms · p90 432 ms

    Logic 41.9%

    Its run record (Method) →

    334 ms432 ms46.0≈ 1.0
  63. SubQ Snap 1.1

    SubQ Snap 1.1 · Subquadratic · decision model

    Parameters: not disclosed by maker

    Shown once with SubQ Snap Preview: the same answer to every question

    Latency: median 381 ms · p90 390 ms

    Logic 52.4%

    Its run record (Method) →

    381 ms390 ms155.7≈ 1.0
  64. celeris-1.1-decision

    celeris-1.1-decision · Celeris · decision model

    Parameters: not disclosed by maker

    Latency: median 381 ms · p90 386 ms

    Logic 47.3%

    Its run record (Method) →

    381 ms386 ms150.5≈ 1.0
  65. celeris-1-decision

    celeris-1-decision · Celeris · decision model

    Parameters: not disclosed by maker

    Latency: median 387 ms · p90 391 ms

    Logic 47.3%

    Its run record (Method) →

    387 ms391 ms153.9≈ 1.0
  66. Perplexity Decider 27B v1.2 API · with web

    Perplexity Decider 27B v1.2 API · with web · Perplexity · decision model

    Parameters: 27B dense

    Latency: median 422 ms · p90 641 ms

    Logic 52.5%

    Its run record (Method) →

    422 ms641 ms105.2≈ 0.9
  67. GLiDE 27B

    GLiDE 27B · Fastino · decision model

    Parameters: 27B dense

    Latency: median 425 ms · p90 461 ms

    Logic 58.4%

    Its run record (Method) →

    425 ms461 ms135.5≈ 1.0
  68. Sieve 31B v3

    Sieve 31B v3 · Sthānika AI · decision model

    Parameters: 31B dense

    Latency: median 637 ms · p90 735 ms

    Logic 57.9%

    Its run record (Method) →

    637 ms735 ms90.6≈ 1.0
  69. KnowLine-31B-Gen1.5

    KnowLine-31B-Gen1.5 · PelaAI · decision model

    Parameters: 31B dense

    Latency: median 764 ms · p90 867 ms

    Logic 59.0%

    Its run record (Method) →

    764 ms867 ms75.6≈ 1.0

Fastino Other OmniJev Liquid AI StartLux PelaAI Perplexity quincyqiang scar-ai Strands Agents newsbang tattrongvu Blockbrain Labs Patrick-18 victorbennett EldanRing vLLM Semantic Router Valen-Team Nace.AI Base Labs Cloudflare Metask Lab Poindexter Labs Apertix AI MachinePulse TypeSafe REAX Logits Google Zefan Cai Turion.AI Inception Microsoft Subquadratic Celeris Sthānika AI 90th percentile

Accuracy across latency

Each setup's accuracy against its median latency (log scale); the dotted line joins the setups no other is both faster and more accurate than; shaded: faster and more accurate than the median setup

1 ms10 ms100 ms1 s10%30%50%70%90%Median latency per question (log scale)AccuracyLower on this axis and more accurate than the median setupPareto front: no other setup is both lower on this axis and more accurate

GLiDE 0.4B: 20.6% at a median 11 ms

GLiClass v4 Edge: 22.7% at a median 12 ms

OneJev1.6-0.8B: 28.2% at a median 15 ms

d1-2B: 37.1% at a median 16 ms

GLiDE 0.8B: 32.1% at a median 20 ms

StartLux-Decision-v1.1-0.8B-Instant: 29.7% at a median 24 ms

KnowLine-0.8B-Gen1: 28.2% at a median 25 ms

GLiDE 2B: 36.8% at a median 28 ms

Perplexity Decider Flash v1.3: 55.9% at a median 28 ms

Chinese Decision Pointer 4B: 50.4% at a median 28 ms

vera-medium-omni: 56.8% at a median 31 ms

Strands Decider 2B: 28.7% at a median 35 ms

StartLux-Decision-v1.1-2B-Instant: 38.7% at a median 36 ms

Adjudex 9B v5: 62.1% at a median 36 ms

Jeffan-9B-1211: 59.8% at a median 37 ms

KnowLine-9B-Gen1: 59.8% at a median 37 ms

Cygnet: 44.3% at a median 37 ms

Perplexity Decider Flash v1.2: 55.2% at a median 40 ms

KnowLine-9B-Gen2: 60.1% at a median 41 ms

Dexy-Decision-9B-V2: 58.8% at a median 45 ms

Verdin-4B-v1.1: 51.5% at a median 47 ms

KnowLine-12B-Gen1: 66.3% at a median 49 ms

KnowLine-2B-Gen1: 35.3% at a median 50 ms

Winnow-12B: 46.9% at a median 52 ms

d3: 70.4% at a median 55 ms

Valen-Decision-1.5-9B: 55.3% at a median 58 ms

StartLux-Decision-v1.1-4B-Instant: 55.8% at a median 67 ms

KnowLine 4B Gen5.5: 56.0% at a median 74 ms

Drex Max: 70.3% at a median 76 ms

GLiDE (v1.2): 69.9% at a median 79 ms

OneJev-27B: 53.8% at a median 79 ms

Elephant-1: 61.3% at a median 80 ms

Perplexity Decider 27B v1.3 (NVIDIA RTX PRO 6000 Blackwell): 72.5% at a median 91 ms

Clef-Flash: 49.8% at a median 99 ms

Metask-Jev-4B: 39.2% at a median 113 ms

KnowLine-31B-Gen2: 80.3% at a median 114 ms

Copernicus-31B: 77.3% at a median 118 ms

d3-ultra: 71.5% at a median 118 ms

Mammoth-1: 74.0% at a median 140 ms

StartLux-Decision-v1.1-9B-Instant: 58.4% at a median 140 ms

Cerno 9B v1.2: 56.0% at a median 153 ms

Karpo Decision 31B: 81.6% at a median 155 ms

Jev 1.14: 65.1% at a median 163 ms

Jev 1.13: 64.1% at a median 166 ms

d1.1: 65.2% at a median 172 ms

GLiDE 9B: 57.9% at a median 174 ms

d1-ultra: 74.1% at a median 179 ms

Clef: 51.1% at a median 181 ms

d1: 55.6% at a median 184 ms

REAX 1.3: 78.0% at a median 189 ms

Logits Sigma-1 Duo: 75.2% at a median 196 ms

d1.2: 65.5% at a median 204 ms

d1.2-ultra: 69.9% at a median 215 ms

Gemini Decision 1 (20261111): 71.3% at a median 226 ms

Perplexity Decider 27B v1.2 (API): 69.8% at a median 226 ms

Open-Jev-27B-v1.1: 54.6% at a median 233 ms

GLiDE (API): 70.8% at a median 264 ms

Turion Jev 31B: 69.0% at a median 278 ms

Mercury Decide: 53.5% at a median 288 ms

Microsoft-Decision-1: 59.4% at a median 310 ms

Drex Max v1.1: 70.1% at a median 326 ms

Drex 2.0: 59.0% at a median 334 ms

SubQ Snap 1.1: 68.6% at a median 381 ms

celeris-1.1-decision: 69.9% at a median 381 ms

celeris-1-decision: 70.0% at a median 387 ms

Perplexity Decider 27B v1.2 (API · with web): 69.3% at a median 422 ms

GLiDE 27B: 72.0% at a median 425 ms

Sieve 31B v3: 77.1% at a median 637 ms

KnowLine-31B-Gen1.5: 77.4% at a median 764 ms

Fastino Other OmniJev Liquid AI StartLux PelaAI Perplexity quincyqiang scar-ai Strands Agents newsbang tattrongvu Blockbrain Labs Patrick-18 victorbennett EldanRing vLLM Semantic Router Valen-Team Nace.AI Base Labs Cloudflare Metask Lab Poindexter Labs Apertix AI MachinePulse TypeSafe REAX Logits Google Zefan Cai Turion.AI Inception Microsoft Subquadratic Celeris Sthānika AI

Latency is the wall time of each request from our machine; runs that shared a GPU or ran with several requests in flight slow each other down, so throughput (questions / min) depends on the run's concurrency. In flight: requests in flight on average, the sum of request latencies over the run's active time (its first to last answer with gaps of more than 10 minutes left out), e.g. “≈ 6” is six at a time.

Median latency is per request; with several requests in flight a run finishes faster than requests × median.

SubQ Snap Preview and SubQ Snap 1.1 returned the same answer to every question (and the same probabilities on 2,151 of 2,163); shown once.

Perplexity Decider 27B v1.2 (API) and Perplexity Decider 27B v1.2 (NVIDIA RTX PRO 6000 Blackwell): two setups of the same model, one on an NVIDIA RTX PRO 6000 Blackwell (96 GB) and one through Perplexity's API; not verified to be identical weights or settings. Its row is the API setup on every test; Perplexity Decider 27B v1.2 (NVIDIA RTX PRO 6000 Blackwell) is under the row's other settings and All setups.

Every setup, every measure

SetupWebAccuracy% of N (N ≤ 2,163) · Wilson 95%Tie groupHolm · McNemarRight / NquestionsKept at ≥99%right / wrongAUROCright vs wrongNon-hallucinationof answers not right, % “I don't know” · higher is betterMedian latencymsp90ms, nearest rankCost$ per 1,000 qper 1M q
Karpo Decision 31B Open · 31B

Karpo Decision 31B · MachinePulse · decision model

Parameters: 31B dense

Latency: median 155 ms · p90 186 ms

Logic 60.7%

Its run record (Method) →

—81.6%79.9%–83.1%a1,764 / 2,163837 right of 837 kept0 mistakes · 1,326 skipped0.9034.8%answered 99%155 ms186 ms$0.113$113estimate · RTX PRO 6000 Blackwell
KnowLine-31B-Gen2 Open · 31B

KnowLine-31B-Gen2 · PelaAI · decision model

Parameters: 31B dense

Base model: Gemma 4 31B

Latency: median 114 ms · p90 120 ms

Logic 58.9%

Its run record (Method) →

—80.3%78.5%–81.9%ab1,736 / 2,163469 right of 470 kept1 wrong · 1,693 skipped0.8980.2%answered 100%114 ms120 ms$0.083$83.14estimate · RTX PRO 6000 Blackwell
REAX 1.3 Open · 57B

REAX 1.3 · REAX · decision model

Parameters: 57.36B dense

Latency: median 189 ms · p90 232 ms

Logic 59.3%

Its run record (Method) →

—78.0%76.2%–79.7%bc1,687 / 2,163178 right of 178 kept0 mistakes · 1,985 skipped0.9032.3%answered 99%189 ms232 ms$0.142$142estimate · RTX PRO 6000 Blackwell
KnowLine-31B-Gen1.5 Open · 31B

KnowLine-31B-Gen1.5 · PelaAI · decision model

Parameters: 31B dense

Latency: median 764 ms · p90 867 ms

Logic 59.0%

Its run record (Method) →

—77.4%75.6%–79.1%cd1,674 / 2,163464 right of 465 kept1 wrong · 1,698 skipped0.8951.1%answered 100%764 ms867 ms$0.047$46.87estimate · DGX Spark
Copernicus-31B Open · 31B

Copernicus-31B · Poindexter Labs · decision model

Parameters: 31B dense

Latency: median 118 ms · p90 142 ms

Logic 58.7%

Its run record (Method) →

—77.3%75.5%–79.1%cd1,673 / 2,163318 right of 318 kept0 mistakes · 1,845 skipped0.8991.7%answered 100%118 ms142 ms$0.084$84.42estimate · RTX PRO 6000 Blackwell
Sieve 31B v3 Open · 31B

Sieve 31B v3 · Sthānika AI · decision model

Parameters: 31B dense

Latency: median 637 ms · p90 735 ms

Logic 57.9%

Its run record (Method) →

—77.1%75.3%–78.8%cd1,668 / 2,163215 right of 215 kept0 mistakes · 1,948 skipped0.8912.1%answered 100%637 ms735 ms$0.039$39.14estimate · DGX Spark
Logits Sigma-1 Duo Open · 57B

Logits Sigma-1 Duo · Logits · decision model

Parameters: 57.36B dense

Latency: median 196 ms · p90 221 ms

Logic 57.8%

Its run record (Method) →

—75.2%73.4%–77.0%de1,627 / 2,16315 right of 15 kept0 mistakes · 2,148 skipped0.8992.1%answered 99%196 ms221 ms$0.141$141estimate · RTX PRO 6000 Blackwell
d1-ultra Closed · Efficient

d1-ultra · Liquid AI · decision model

Parameters: not disclosed by maker (built on a larger open-weight model, not named with a count)

Latency: median 179 ms · p90 426 ms

Logic 54.9%

Its run record (Method) →

—74.1%72.2%–75.9%defg1,603 / 2,163213 right of 213 kept0 mistakes · 1,950 skipped0.87687.4%answered 61%179 ms426 ms$0.006$6.14
Mammoth-1 Open · 27B

Mammoth-1 · Base Labs · decision model

Parameters: 27B dense

Latency: median 140 ms · p90 173 ms

Logic 54.5%

Its run record (Method) →

—74.0%72.1%–75.8%defh1,600 / 2,163412 right of 412 kept0 mistakes · 1,751 skipped0.89570.6%answered 75%140 ms173 ms$0.023$22.56 list$0.104$104estimate · RTX PRO 6000 Blackwell
Perplexity Decider 27B v1.3 NVIDIA RTX PRO 6000 Blackwell Open · 27B

Perplexity Decider 27B v1.3 NVIDIA RTX PRO 6000 Blackwell · Perplexity · decision model

Parameters: 27B dense

Latency: median 91 ms · p90 103 ms

Logic 54.8%

Its run record (Method) →

—72.5%70.6%–74.3%efghi1,568 / 2,163323 right of 323 kept0 mistakes · 1,840 skipped0.8569.4%answered 97%91 ms103 ms$0.066$65.87estimate · RTX PRO 6000 Blackwell
GLiDE 27B Open · 27B

GLiDE 27B · Fastino · decision model

Parameters: 27B dense

Latency: median 425 ms · p90 461 ms

Logic 58.4%

Its run record (Method) →

—72.0%70.1%–73.8%efghij1,557 / 2,163198 right of 199 kept1 wrong · 1,964 skipped0.86421.6%answered 94%425 ms461 ms$0.026$26.16estimate · DGX Spark
d3-ultra Open · 26B

d3-ultra · vLLM Semantic Router · decision model

Parameters: 26.09B dense

Latency: median 118 ms · p90 148 ms

Logic 54.6%

Its run record (Method) →

—71.5%69.6%–73.4%fghij1,547 / 2,16311 right of 11 kept0 mistakes · 2,152 skipped0.8575.3%answered 98%118 ms148 ms$0.089$88.58estimate · RTX PRO 6000 Blackwell
Gemini Decision 1 20261111 Closed · Efficient

Gemini Decision 1 20261111 · Google · decision model

Parameters: not disclosed by maker (closed weights)

Latency: median 226 ms · p90 351 ms

Logic 50.0%

Its run record (Method) →

—71.3%69.4%–73.2%efghij1,543 / 2,163214 right of 214 kept0 mistakes · 1,949 skipped0.8593.9%answered 99%226 ms351 ms$0.022$22.02
GLiDE API Closed · Efficient

GLiDE API · Fastino · decision model

Parameters: not disclosed by maker

Latency: median 264 ms · p90 279 ms

Logic 54.5%

Its run record (Method) →

—70.8%68.8%–72.7%fghij1,531 / 2,163161 right of 161 kept0 mistakes · 2,002 skipped0.86540.6%answered 86%264 ms279 ms$0.028$28.30
d3 Open · 26B

d3 · vLLM Semantic Router · decision model

Parameters: 26.09B dense

Latency: median 55 ms · p90 62 ms

Logic 54.5%

Its run record (Method) →

—70.4%68.5%–72.3%gij1,523 / 2,163162 right of 163 kept1 wrong · 2,000 skipped0.8784.7%answered 99%55 ms62 ms$0.041$40.61estimate · RTX PRO 6000 Blackwell
Drex Max Open · 27B

Drex Max · Nace.AI · decision model

Parameters: 27B dense

Latency: median 76 ms · p90 89 ms

Logic 53.1%

Its run record (Method) →

—70.3%68.4%–72.2%gijk1,521 / 2,163719 right of 729 kept10 wrong · 1,434 skipped0.8614.5%answered 99%76 ms89 ms$0.013$13.06 list$0.055$54.68estimate · RTX PRO 6000 Blackwell
Drex Max v1.1 Closed · Efficient (provisional)

Drex Max v1.1 · Nace.AI · decision model

Parameters: not disclosed by maker

Latency: median 326 ms · p90 473 ms

Logic 54.8%

Its run record (Method) →

—70.1%68.1%–72.0%ijk1,516 / 2,163315 right of 315 kept0 mistakes · 1,848 skipped0.86743.5%answered 84%326 ms473 ms$0.013$12.88
celeris-1-decision Closed · Efficient

celeris-1-decision · Celeris · decision model

Parameters: not disclosed by maker

Latency: median 387 ms · p90 391 ms

Logic 47.3%

Its run record (Method) →

—70.0%68.0%–71.9%hijk1,514 / 2,163No answers at this threshold0.85882.7%answered 62%387 ms391 ms$0.009$9.10
GLiDE v1.2 Open · 27B

GLiDE v1.2 · Fastino · decision model

Parameters: 27B dense

Latency: median 79 ms · p90 84 ms

Logic 55.2%

Its run record (Method) →

—69.9%67.9%–71.8%ijk1,511 / 2,163131 right of 131 kept0 mistakes · 2,032 skipped0.86235.8%answered 88%79 ms84 ms$0.008$7.55 list$0.058$57.88estimate · RTX PRO 6000 Blackwell
celeris-1.1-decision Closed · Efficient

celeris-1.1-decision · Celeris · decision model

Parameters: not disclosed by maker

Latency: median 381 ms · p90 386 ms

Logic 47.3%

Its run record (Method) →

—69.9%67.9%–71.8%ijkl1,511 / 2,163No answers at this threshold0.85982.9%answered 62%381 ms386 ms$0.001$0.91
d1.2-ultra Closed · Undisclosed

d1.2-ultra · Liquid AI · decision model

Parameters: not disclosed by maker

Latency: median 215 ms · p90 347 ms

Logic 51.6%

Its run record (Method) →

—69.9%67.9%–71.8%ijk1,511 / 2,163115 right of 115 kept0 mistakes · 2,048 skipped0.87686.7%answered 58%215 ms347 ms$0.006$6.14
Perplexity Decider 27B v1.2 API Open · 27B

Perplexity Decider 27B v1.2 API · Perplexity · decision model

Parameters: 27B dense

Latency: median 226 ms · p90 368 ms

Logic 53.3%

Its run record (Method) →

—69.8%67.8%–71.7%ijkl1,509 / 2,163331 right of 331 kept0 mistakes · 1,832 skipped0.8697.0%answered 98%226 ms368 ms$0.004$4.29
Perplexity Decider 27B v1.2 API · with web Open · 27B

Perplexity Decider 27B v1.2 API · with web · Perplexity · decision model

Parameters: 27B dense

Latency: median 422 ms · p90 641 ms

Logic 52.5%

Its run record (Method) →

with web69.3%67.3%–71.2%jklm1,499 / 2,163326 right of 326 kept0 mistakes · 1,837 skipped0.87315.4%answered 95%422 ms641 ms$1.023$1,023
Turion Jev 31B Open · 31B

Turion Jev 31B · Turion.AI · decision model

Parameters: 31B dense

Latency: median 278 ms · p90 329 ms

Logic 52.3%

Its run record (Method) →

—69.0%67.0%–70.9%ijklmn1,493 / 2,16343 right of 43 kept0 mistakes · 2,120 skipped0.8642.2%answered 99%278 ms329 ms$0.017$17.46estimate · DGX Spark
SubQ Snap 1.1 Closed · Efficient

SubQ Snap 1.1 · Subquadratic · decision model

Parameters: not disclosed by maker

Shown once with SubQ Snap Preview: the same answer to every question

Latency: median 381 ms · p90 390 ms

Logic 52.4%

Its run record (Method) →

—68.6%66.6%–70.5%jklmn1,483 / 2,163360 right of 361 kept1 wrong · 1,802 skipped0.83952.5%answered 80%381 ms390 ms$0.009$8.59
KnowLine-12B-Gen1 Open · 12B

KnowLine-12B-Gen1 · PelaAI · decision model

Parameters: 12B dense

Base model: Gemma 4 12B

Latency: median 49 ms · p90 55 ms

Logic 50.6%

Its run record (Method) →

—66.3%64.2%–68.2%klmno1,433 / 2,1637 right of 7 kept0 mistakes · 2,156 skipped0.8191.1%answered 100%49 ms55 ms$0.036$36.13estimate · RTX PRO 6000 Blackwell
d1.2 Closed · Undisclosed

d1.2 · Liquid AI · decision model

Parameters: not disclosed by maker

Latency: median 204 ms · p90 386 ms

Logic 49.2%

Its run record (Method) →

—65.5%63.5%–67.5%lmnop1,417 / 2,163141 right of 141 kept0 mistakes · 2,022 skipped0.83856.3%answered 76%204 ms386 ms$0.003$2.92
d1.1 Closed · Efficient

d1.1 · Liquid AI · decision model

Parameters: not disclosed by maker

Latency: median 172 ms · p90 290 ms

Logic 46.0%

Its run record (Method) →

—65.2%63.2%–67.2%mnop1,410 / 2,163183 right of 183 kept0 mistakes · 1,980 skipped0.82982.5%answered 56%172 ms290 ms$0.003$2.92
Jev 1.14 Closed · Efficient

Jev 1.14 · TypeSafe · decision model

Parameters: not disclosed by maker

Latency: median 163 ms · p90 185 ms

Logic 48.2%

Its run record (Method) →

—65.1%63.1%–67.1%nop1,409 / 2,163202 right of 202 kept0 mistakes · 1,961 skipped0.8297.7%answered 97%163 ms185 ms$0.019$18.99
Jev 1.13 Closed · Efficient

Jev 1.13 · TypeSafe · decision model

Parameters: not disclosed by maker

Latency: median 166 ms · p90 206 ms

Logic 46.3%

Its run record (Method) →

—64.1%62.0%–66.1%opq1,386 / 2,163248 right of 248 kept0 mistakes · 1,915 skipped0.8345.7%answered 98%166 ms206 ms$0.019$18.99
Adjudex 9B v5 Open · 9.0B

Adjudex 9B v5 · newsbang · decision model

Parameters: 9B dense

Base model: Qwen3.5-9B

Latency: median 36 ms · p90 44 ms

Logic 46.4%

Its run record (Method) →

—62.1%60.0%–64.1%opqr1,343 / 2,163197 right of 198 kept1 wrong · 1,965 skipped0.8063.1%answered 99%36 ms44 ms$0.027$26.54estimate · RTX PRO 6000 Blackwell
Elephant-1 Open · 9.0B

Elephant-1 · Base Labs · decision model

Parameters: 9B dense

Latency: median 80 ms · p90 92 ms

Logic 46.4%

Its run record (Method) →

—61.3%59.3%–63.4%pqr1,327 / 2,163182 right of 183 kept1 wrong · 1,980 skipped0.80365.0%answered 67%80 ms92 ms$0.008$7.90 list$0.058$58.20estimate · RTX PRO 6000 Blackwell
KnowLine-9B-Gen2 Open · 9.0B

KnowLine-9B-Gen2 · PelaAI · decision model

Parameters: 9B dense

Base model: Qwen3.5-9B

Latency: median 41 ms · p90 46 ms

Logic 43.9%

Its run record (Method) →

—60.1%58.0%–62.1%qrs1,300 / 2,163No answers at this threshold0.8033.2%answered 99%41 ms46 ms$0.031$31.02estimate · RTX PRO 6000 Blackwell
Jeffan-9B-1211 Open · 9.0B

Jeffan-9B-1211 · tattrongvu · decision model

Parameters: 9B dense

Base model: Qwen3.5-9B

Latency: median 37 ms · p90 51 ms

Logic 48.2%

Its run record (Method) →

—59.8%57.7%–61.9%qrs1,294 / 2,16364 right of 64 kept0 mistakes · 2,099 skipped0.8146.6%answered 97%37 ms51 ms$0.029$28.78estimate · RTX PRO 6000 Blackwell
KnowLine-9B-Gen1 Open · 9.0B

KnowLine-9B-Gen1 · PelaAI · decision model

Parameters: 9B dense

Base model: Qwen3.5-9B

Latency: median 37 ms · p90 43 ms

Logic 44.3%

Its run record (Method) →

—59.8%57.7%–61.8%qrs1,293 / 2,163123 right of 123 kept0 mistakes · 2,040 skipped0.7990.9%answered 100%37 ms43 ms$0.027$26.54estimate · RTX PRO 6000 Blackwell
Microsoft-Decision-1 Closed · Efficient

Microsoft-Decision-1 · Microsoft · decision model

Parameters: not disclosed by maker

Base model: Qwen3.5-9B

Latency: median 310 ms · p90 359 ms

Logic 44.4%

Its run record (Method) →

—59.4%57.3%–61.4%rst1,284 / 2,16327 right of 27 kept0 mistakes · 2,136 skipped0.7875.1%answered 98%310 ms359 ms$0.007$6.81
Drex 2.0 Open · 35B MoE · 3.0B active

Drex 2.0 · Nace.AI · decision model

Parameters: 35B MoE · 3B active

Latency: median 334 ms · p90 432 ms

Logic 41.9%

Its run record (Method) →

—59.0%56.9%–61.0%rstu1,276 / 2,163174 right of 174 kept0 mistakes · 1,989 skipped0.81320.7%answered 91%334 ms432 ms$0.009$8.92
Dexy-Decision-9B-V2 Open · 9.0B

Dexy-Decision-9B-V2 · Patrick-18 · decision model

Parameters: 9B dense

Base model: vllm-sr/d3-flash

Latency: median 45 ms · p90 50 ms

Logic 45.3%

Its run record (Method) →

—58.8%56.7%–60.9%rstu1,272 / 2,163207 right of 210 kept3 wrong · 1,953 skipped0.8041.7%answered 99%45 ms50 ms$0.032$31.98estimate · RTX PRO 6000 Blackwell
StartLux-Decision-v1.1-9B-Instant Open · 9.0B

StartLux-Decision-v1.1-9B-Instant · StartLux · decision model

Parameters: 9B dense

Latency: median 140 ms · p90 146 ms

Logic 46.6%

Its run record (Method) →

—58.4%56.3%–60.5%rstuv1,264 / 2,16349 right of 49 kept0 mistakes · 2,114 skipped0.7814.2%answered 98%140 ms146 ms$0.008$8.34estimate · DGX Spark
GLiDE 9B Open · 7.9B

GLiDE 9B · Fastino · decision model

Parameters: 7.94B dense

Latency: median 174 ms · p90 182 ms

Logic 47.8%

Its run record (Method) →

—57.9%55.8%–60.0%rstuvw1,253 / 2,16373 right of 73 kept0 mistakes · 2,090 skipped0.79824.6%answered 89%174 ms182 ms$0.010$10.49estimate · DGX Spark
vera-medium-omni Open · 4.2B

vera-medium-omni · scar-ai · decision model

Parameters: 4.2B dense

Base model: Qwen3.5-4B

Latency: median 31 ms · p90 35 ms

Logic 44.1%

Its run record (Method) →

—56.8%54.7%–58.9%stuvw1,229 / 2,16333 right of 33 kept0 mistakes · 2,130 skipped0.7893.8%answered 98%31 ms35 ms$0.023$23.34estimate · RTX PRO 6000 Blackwell
Cerno 9B v1.2 Open · 9.0B

Cerno 9B v1.2 · Apertix AI · decision model

Parameters: 9B dense

Latency: median 153 ms · p90 175 ms

Logic 43.3%

Its run record (Method) →

—56.0%53.9%–58.1%stuvw1,212 / 2,163244 right of 254 kept10 wrong · 1,909 skipped0.7801.2%answered 99%153 ms175 ms$0.010$9.51estimate · DGX Spark
KnowLine 4B Gen5.5 Open · 4.0B

KnowLine 4B Gen5.5 · PelaAI · decision model

Parameters: 4B dense

Latency: median 74 ms · p90 84 ms

Logic 42.3%

Its run record (Method) →

—56.0%53.9%–58.1%stuvw1,211 / 2,16354 right of 54 kept0 mistakes · 2,109 skipped0.7651.6%answered 99%74 ms84 ms$0.005$4.56estimate · DGX Spark
Perplexity Decider Flash v1.3 Open · 4.0B

Perplexity Decider Flash v1.3 · Perplexity · decision model

Parameters: 4B dense

Latency: median 28 ms · p90 31 ms

Logic 44.7%

Its run record (Method) →

—55.9%53.8%–58.0%stuvw1,210 / 2,163113 right of 114 kept1 wrong · 2,049 skipped0.7878.4%answered 96%28 ms31 ms$0.022$22.38estimate · RTX PRO 6000 Blackwell
StartLux-Decision-v1.1-4B-Instant Open · 4.0B

StartLux-Decision-v1.1-4B-Instant · StartLux · decision model

Parameters: 4B dense

Latency: median 67 ms · p90 69 ms

Logic 45.9%

Its run record (Method) →

—55.8%53.7%–57.9%stuvwx1,208 / 2,16396 right of 96 kept0 mistakes · 2,067 skipped0.77913.4%answered 94%67 ms69 ms$0.004$4.10estimate · DGX Spark
d1 Closed · Efficient

d1 · Liquid AI · decision model

Parameters: not disclosed by maker

Latency: median 184 ms · p90 297 ms

Logic 40.8%

Its run record (Method) →

—55.6%53.5%–57.7%stuvwxy1,203 / 2,163152 right of 152 kept0 mistakes · 2,011 skipped0.7672.9%answered 99%184 ms297 ms$0.003$2.92
Valen-Decision-1.5-9B Open · 9.0B

Valen-Decision-1.5-9B · Valen-Team · decision model

Parameters: 9B dense

Base model: Qwen3.5-9B

Latency: median 58 ms · p90 64 ms

Logic 44.4%

Its run record (Method) →

—55.3%53.2%–57.4%tuvwxy1,197 / 2,163130 right of 130 kept0 mistakes · 2,033 skipped0.7902.6%answered 99%58 ms64 ms$0.041$40.93estimate · RTX PRO 6000 Blackwell
Perplexity Decider Flash v1.2 Open · 4.0B

Perplexity Decider Flash v1.2 · Perplexity · decision model

Parameters: 4B dense

Latency: median 40 ms · p90 47 ms

Logic 43.1%

Its run record (Method) →

—55.2%53.1%–57.3%tuvwxy1,195 / 2,163150 right of 155 kept5 wrong · 2,008 skipped0.78112.0%answered 94%40 ms47 ms$0.002$2.18 list$0.030$30.06estimate · RTX PRO 6000 Blackwell
Open-Jev-27B-v1.1 Open · 27B

Open-Jev-27B-v1.1 · Zefan Cai · decision model

Parameters: 27B dense

Base model: Qwen3.8-27B + adapter

Latency: median 233 ms · p90 272 ms

Logic 42.6%

Its run record (Method) →

—54.6%52.4%–56.6%uvwxyz1,180 / 2,163No answers at this threshold0.7645.0%answered 98%233 ms272 ms$0.169$169estimate · RTX PRO 6000 Blackwell
OneJev-27B Open · 27B

OneJev-27B · OmniJev · decision model

Parameters: 27B dense

Latency: median 79 ms · p90 106 ms

Logic 40.6%

Its run record (Method) →

—53.8%51.7%–55.9%vwxyzA1,163 / 2,16326 right of 26 kept0 mistakes · 2,137 skipped0.7831.8%answered 99%79 ms106 ms$0.061$60.76estimate · RTX PRO 6000 Blackwell
Mercury Decide Closed · Efficient

Mercury Decide · Inception · decision model

Parameters: not disclosed by maker

Latency: median 288 ms · p90 342 ms

Logic 45.1%

Its run record (Method) →

—53.5%51.4%–55.6%wxyzA1,157 / 2,163272 right of 281 kept9 wrong · 1,882 skipped0.7893.2%answered 99%288 ms342 ms$0.004$4.38
Verdin-4B-v1.1 Open · 4.0B

Verdin-4B-v1.1 · victorbennett · decision model

Parameters: 4B dense

Base model: Qwen3.5-4B

Latency: median 47 ms · p90 52 ms

Logic 39.7%

Its run record (Method) →

—51.5%49.4%–53.6%yzAB1,114 / 2,163126 right of 130 kept4 wrong · 2,033 skipped0.7526.4%answered 97%47 ms52 ms$0.034$33.58estimate · RTX PRO 6000 Blackwell
Clef Open · 27B

Clef · Cloudflare · decision model

Parameters: 27B dense

Latency: median 181 ms · p90 210 ms

Logic 40.8%

Its run record (Method) →

—51.1%49.0%–53.2%xyzAB1,105 / 2,16312 right of 12 kept0 mistakes · 2,151 skipped0.7671.6%answered 99%181 ms210 ms$0.069$68.54 list$0.131$131estimate · RTX PRO 6000 Blackwell
Chinese Decision Pointer 4B Open · 4.0B

Chinese Decision Pointer 4B · quincyqiang · decision model

Parameters: 4B dense

Base model: Qwen3.5-4B

Latency: median 28 ms · p90 32 ms

Logic 40.4%

Its run record (Method) →

—50.4%48.3%–52.5%zAB1,091 / 2,163149 right of 155 kept6 wrong · 2,008 skipped0.7463.4%answered 98%28 ms32 ms$0.023$22.70estimate · RTX PRO 6000 Blackwell
Clef-Flash Open · 9.0B

Clef-Flash · Cloudflare · decision model

Parameters: 9B dense

Latency: median 99 ms · p90 127 ms

Logic 38.9%

Its run record (Method) →

—49.8%47.7%–51.9%AB1,077 / 2,163No answers at this threshold0.7302.0%answered 99%99 ms127 ms$0.011$10.85 list$0.072$71.95estimate · RTX PRO 6000 Blackwell
Winnow-12B Open · 12B

Winnow-12B · EldanRing · decision model

Parameters: 12B dense

Latency: median 52 ms · p90 61 ms

Logic 39.7%

Its run record (Method) →

—46.9%44.8%–49.0%BC1,014 / 2,16320 right of 21 kept1 wrong · 2,142 skipped0.7439.2%answered 95%52 ms61 ms$0.038$38.37estimate · RTX PRO 6000 Blackwell
Cygnet Open · 12B

Cygnet · Blockbrain Labs · decision model

Parameters: 11.95B dense

Base model: Gemma 4 12B IT

Latency: median 37 ms · p90 41 ms

Logic 39.3%

Its run record (Method) →

—44.3%42.3%–46.4%C959 / 2,163No answers at this threshold0.7354.9%answered 97%37 ms41 ms$0.024$23.94 list$0.028$27.50estimate · RTX PRO 6000 Blackwell
Metask-Jev-4B Open · 4.5B

Metask-Jev-4B · Metask Lab · decision model

Parameters: 4.54B dense

Latency: median 113 ms · p90 144 ms

Logic 32.1%

Its run record (Method) →

—39.2%37.1%–41.2%D847 / 2,163No answers at this threshold0.6512.2%answered 99%113 ms144 ms$0.083$83.14estimate · RTX PRO 6000 Blackwell
StartLux-Decision-v1.1-2B-Instant Open · 2.0B

StartLux-Decision-v1.1-2B-Instant · StartLux · decision model

Parameters: 2B dense

Latency: median 36 ms · p90 37 ms

Logic 30.8%

Its run record (Method) →

—38.7%36.6%–40.7%D836 / 2,1632 right of 2 kept0 mistakes · 2,161 skipped0.6857.2%answered 95%36 ms37 ms$0.002$2.27estimate · DGX Spark
d1-2B Open · 1.9B

d1-2B · Liquid AI · decision model

Parameters: 1.92B dense

Base model: LFM3-VL-2B

Latency: median 16 ms · p90 21 ms

Logic 30.7%

Its run record (Method) →

—37.1%35.1%–39.1%D802 / 2,1634 right of 4 kept0 mistakes · 2,159 skipped0.65680.2%answered 34%16 ms21 ms$0.014$14.07estimate · RTX PRO 6000 Blackwell
GLiDE 2B Open · 1.9B

GLiDE 2B · Fastino · decision model

Parameters: 1.88B dense

Latency: median 28 ms · p90 30 ms

Logic 32.3%

Its run record (Method) →

—36.8%34.7%–38.8%D795 / 2,16315 right of 16 kept1 wrong · 2,147 skipped0.63517.3%answered 89%28 ms30 ms$0.002$1.78estimate · DGX Spark
KnowLine-2B-Gen1 Open · 2.0B

KnowLine-2B-Gen1 · PelaAI · decision model

Parameters: 2B dense

Base model: Qwen3.5-2B

Latency: median 50 ms · p90 54 ms

Logic 31.7%

Its run record (Method) →

—35.3%33.3%–37.3%DE763 / 2,1638 right of 8 kept0 mistakes · 2,155 skipped0.6361.7%answered 99%50 ms54 ms$0.003$3.20estimate · DGX Spark
GLiDE 0.8B Open · 0.8B

GLiDE 0.8B · Fastino · decision model

Parameters: 0.75B dense

Latency: median 20 ms · p90 21 ms

Logic 29.2%

Its run record (Method) →

—32.1%30.2%–34.1%EF694 / 2,1634 right of 4 kept0 mistakes · 2,159 skipped0.58920.8%answered 85%20 ms21 ms$0.001$1.28estimate · DGX Spark
StartLux-Decision-v1.1-0.8B-Instant Open · 0.8B

StartLux-Decision-v1.1-0.8B-Instant · StartLux · decision model

Parameters: 0.8B dense

Base model: Qwen3.5-0.8B

Latency: median 24 ms · p90 25 ms

Logic 24.7%

Its run record (Method) →

—29.7%27.8%–31.6%F642 / 2,163No answers at this threshold0.58311.3%answered 92%24 ms25 ms$0.002$1.53estimate · DGX Spark
Strands Decider 2B Open · 2.0B

Strands Decider 2B · Strands Agents · decision model

Parameters: 2B dense

Latency: median 35 ms · p90 41 ms

Logic 22.4%

Its run record (Method) →

—28.7%26.8%–30.6%F620 / 2,163No answers at this threshold0.5693.2%answered 98%35 ms41 ms$0.026$26.22estimate · RTX PRO 6000 Blackwell
OneJev1.6-0.8B Open · 0.8B

OneJev1.6-0.8B · OmniJev · decision model

Parameters: 0.8B dense

Latency: median 15 ms · p90 17 ms

Logic 27.0%

Its run record (Method) →

—28.2%26.4%–30.2%F611 / 2,163No answers at this threshold0.5920.6%answered 100%15 ms17 ms$0.001$1.09estimate · DGX Spark
KnowLine-0.8B-Gen1 Open · 0.8B

KnowLine-0.8B-Gen1 · PelaAI · decision model

Parameters: 0.8B dense

Base model: Qwen3.5-0.8B

Latency: median 25 ms · p90 28 ms

Logic 26.7%

Its run record (Method) →

—28.2%26.3%–30.1%F610 / 2,163No answers at this threshold0.5960.3%answered 100%25 ms28 ms$0.002$1.69estimate · DGX Spark
GLiClass v4 Edge Open · 0.4B

GLiClass v4 Edge · decision model

Parameters: 0.4B dense

Base model: DeBERTa-v3-large

Latency: median 12 ms · p90 13 ms

Logic 23.6%

Its run record (Method) →

—22.7%20.9%–24.5%G490 / 2,163No answers at this threshold0.50117.2%answered 86%12 ms13 ms$0.001$0.85estimate · DGX Spark
GLiDE 0.4B Open · 0.4B

GLiDE 0.4B · Fastino · decision model

Parameters: 0.40B dense

Latency: median 11 ms · p90 12 ms

Logic 23.2%

Its run record (Method) →

—20.6%18.9%–22.3%G445 / 2,163No answers at this threshold0.50072.8%answered 33%11 ms12 ms$0.001$0.77estimate · DGX Spark

Tie groups and “not measured”

estimate · RTX PRO 6000 Blackwell / DGX Spark = run on our own hardware; its cost is an estimate, explained on hover, focus or tap of the amount: the machine's wall-clock time on the run, first to last answer while it was running (pauses, estimated from gaps of more than 10 minutes between answers, left out and listed), × $2.49/h per NVIDIA RTX PRO 6000 Blackwell (96 GB) (Runpod's price for the card) or × $0.213/h per NVIDIA DGX Spark (GB10) (its purchase over three years plus electricity; see Method), each run charged in full even when runs shared the machine, so at most; sorting and the cost chart use the estimate; list = maker's list price × our token counts; unmarked = provider's price × usage the API reported.