Reference · Models

The lineup, at the token level.

  1. GPT-5.5 Azure OpenAI
    S G P D
    Premium · hardest only · keep narrow
    $5.00 / $30.00
    5%
  2. Claude Opus 4.7 Bedrock
    S G P D
    Premium · hardest queries · ambiguous multi-doc
    $5.00 / $25.00
    12%
  3. Gemini 3 Google
    S G P D
    Multimodal · chart synthesis · screenshot reasoning
    $3.50 / $15.00
    5%
  4. Claude Sonnet 4.6 Bedrock
    S G P D
    Default · English reports · charts · the workhorse
    $3.00 / $15.00
    46%
  5. DeepSeek V4 Pro Alibaba
    S G P D
    Default + math + long-context · the bet of this plan
    $1.74 / $3.48
    35%
  6. Qwen 3.6 Max Alibaba
    S G P D
    Chinese frontier · mainland-specific only
    $1.30 / $7.80
    10%
  7. GPT-5 Azure OpenAI
    S G P D
    Default · English reports · charts
    $1.25 / $10.00
    17%
  8. o4-mini Azure OpenAI
    S G P D
    Agentic chains · OpenAI tool format
    $1.10 / $4.40
    10%
  9. Claude Haiku 4.5 Bedrock
    S G P D
    Fast English lookups · cheap path · classification
    $1.00 / $5.00
    18%
  10. Gemma 4 Google
    S G P D
    Open weights · multilingual utility · keeps Google in the mix
    $0.40 / $1.20
    5%
  11. GPT-5 mini Azure OpenAI
    S G P D
    English lookups · structured-output reliability
    $0.25 / $2.00
    15%
  12. DeepSeek V4 Flash Alibaba
    S G P D
    Bulk cheap path · multilingual utility
    $0.14 / $0.28
    20%

May 2026 list price · cache-aware blended cost. S · G · P · D = Silver · Gold · Platinum · Diamond — the highlighted letter is the currently active plan. Routing % is the share of queries this lane wins under that plan.