Skip to main content

Benchmarks: True Prices vs. other AIs

Building markets is a constant cat-and-mouse game between sellers who want more of your money, buyers who want to spend less of their money, and platforms, including search/AI platforms, that want to extract as much as possible from everybody involved and make as much profit as possible. We update these benchmarks periodically.

How do you know if you actually have the best price? Who is the referee? How do you understand total-delivered cost, or all the nuances of every market from eligibility to availability? Does trust factor in? Reliability? Understanding how to deal with poor, misleading, intentionally sparse, estimated, incorrect, poisoned, or overwhelming data? What about when information is hidden or dynamic, incorrectly recorded, or anomalous? Intentionally hidden by stores or tightly regulated? What about functional equivalents? Fees can be hidden until the last minute, or split up or split under different names but actually be required, or have different conditions. Predatory terms? What about markets that factor in radiuses and jurisdictions? Confusing or contradictory prices and information? What about conditions you only become aware of at checkout, from local rules to obsolete or inapplicable modifiers? Foreign and domestic counterparties? Trust, and signaling? Variants and terminology, or information that can only be understood via images? True Prices specializes in all of this, and these benchmarks are mindful of those nuances and many more.

Every day we fight a battle against all those who don't want their prices known. Because when prices are known, people know what to choose. And that means they'll choose better deals.

E-commerce markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -67.8%
    Average savings (when looking for best prices): -67.8%
    % of time that best price found when actively searching for it: 92%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 1.5%

    True Pricesmax

  • Cost per message$1.81124.7s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): 37.8%
    Average savings (when looking for best prices): -25.1%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 39%

    MetaMuse Wonder (max + best tools)

  • Cost per message$1.96247.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): 65.2%
    Average savings (when looking for best prices): -21.6%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 37.3%

    Grok4.7 (max + best tools)

  • Cost per message$0.46176.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): 34.7%
    Average savings (when looking for best prices): -16.5%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 59.8%

    GoogleArgon (max + best tools)

  • Cost per message$2.46393.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): 33.8%
    Average savings (when looking for best prices): -29.9%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 35.4%

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.78308.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): 73.5%
    Average savings (when looking for best prices): -35.5%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 51.8%

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.65483.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): 112.7%
    Average savings (when looking for best prices): -17.6%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 56.2%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.14192.4s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): 64%
    Average savings (when looking for best prices): -37.8%
    % of time that best price found when actively searching for it: 8%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 47.2%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.21370.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): 74%
    Average savings (when looking for best prices): -25.3%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 52.5%

    MistralLarge 4 (max + best tools)

Fuel markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -21.4%
    Average savings (when looking for best prices): -21.4%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 2%

    True Pricesmax

  • Cost per message$2.24166.7s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MetaMuse Wonder (max + best tools)

  • Cost per message$1.96158.1s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    Grok4.7 (max + best tools)

  • Cost per message$0.52225.1s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    GoogleArgon (max + best tools)

  • Cost per message$2.81337.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    OpenAIAstra Pro (max + best tools)

  • Cost per message$2.20416.7s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.64255.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): -2.3%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 97.6%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.18346.7s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.24377.0s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MistralLarge 4 (max + best tools)

Rideshare markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -26.7%
    Average savings (when looking for best prices): -26.7%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 4.6%

    True Pricesmax

  • Cost per message$1.82103.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MetaMuse Wonder (max + best tools)

  • Cost per message$1.96140.7s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    Grok4.7 (max + best tools)

  • Cost per message$0.73600.0s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    GoogleArgon (max + best tools)

  • Cost per message$2.96270.5s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.44304.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.76371.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.17365.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.16157.7s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MistralLarge 4 (max + best tools)

Hotels markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -36.4%
    Average savings (when looking for best prices): -36.4%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 2.3%

    True Pricesmax

  • Cost per message$1.72165.7s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MetaMuse Wonder (max + best tools)

  • Cost per message$1.96149.4s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    Grok4.7 (max + best tools)

  • Cost per message$0.52153.7s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    GoogleArgon (max + best tools)

  • Cost per message$2.09248.8s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.56283.3s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.66267.1s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.15299.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.18291.4s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MistralLarge 4 (max + best tools)

Flights markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -11.4%
    Average savings (when looking for best prices): -11.4%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 0%

    True Pricesmax

  • Cost per message$2.06144.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): 12.8%
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 50%

    MetaMuse Wonder (max + best tools)

  • Cost per message$1.96174.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    Grok4.7 (max + best tools)

  • Cost per message$0.53261.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): 5.3%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 50%

    GoogleArgon (max + best tools)

  • Cost per message$2.61292.4s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.89338.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$1.02386.0s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.22600.0s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): API unavailable
    % of time that best price found when actively searching for it: API unavailable
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.14147.4s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MistralLarge 4 (max + best tools)

Car rentals markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -43.7%
    Average savings (when looking for best prices): -43.7%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 0%

    True Pricesmax

  • Cost per message$1.84125.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    MetaMuse Wonder (max + best tools)

  • Cost per message$1.96146.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    Grok4.7 (max + best tools)

  • Cost per message$0.47216.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    GoogleArgon (max + best tools)

  • Cost per message$2.47238.0s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.77393.8s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.67234.0s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.19382.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.20246.1s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    MistralLarge 4 (max + best tools)

Grocery basket markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -44.7%
    Average savings (when looking for best prices): -44.7%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 7.1%

    True Pricesmax

  • Cost per message$1.78152.6s
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MetaMuse Wonder (max + best tools)

  • Cost per message$1.96142.4s
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    Grok4.7 (max + best tools)

  • Cost per message$0.41240.2s
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    GoogleArgon (max + best tools)

  • Cost per message$2.53283.9s
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.46298.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -14.3%
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 83.3%

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.52241.6s
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.14275.9s
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.24373.0s
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MistralLarge 4 (max + best tools)

Wholesale food markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -68.2%
    Average savings (when looking for best prices): -68.2%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 0%

    True Pricesmax

  • Cost per message$2.58234.4s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -34.8%
    Average savings (when looking for best prices): -68.1%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 26.9%

    MetaMuse Wonder (max + best tools)

  • Cost per message$0.53286.3s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -47.1%
    Average savings (when looking for best prices): -68.1%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 21.4%

    Grok4.7 (max + best tools)

  • Cost per message$0.71255.1s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -47.1%
    Average savings (when looking for best prices): -59.5%
    % of time that best price found when actively searching for it: 71.4%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 47.2%

    GoogleArgon (max + best tools)

  • Cost per message$2.53325.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -47.1%
    Average savings (when looking for best prices): -54.9%
    % of time that best price found when actively searching for it: 71.4%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 25%

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.73420.0s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -47.1%
    Average savings (when looking for best prices): -53.6%
    % of time that best price found when actively searching for it: 71.4%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 43.8%

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$1.07402.8s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -23.5%
    Average savings (when looking for best prices): -36.7%
    % of time that best price found when actively searching for it: 42.9%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 85.8%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.14592.7s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -47.1%
    Average savings (when looking for best prices): -68.1%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 30%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.06256.3s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused prices

    MistralLarge 4 (max + best tools)

Medicines markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -91%
    Average savings (when looking for best prices): -91%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 0%

    True Pricesmax

  • Cost per message$1.24117.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MetaMuse Wonder (max + best tools)

  • Cost per message$1.96166.7s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    Grok4.7 (max + best tools)

  • Cost per message$0.51207.1s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    GoogleArgon (max + best tools)

  • Cost per message$1.84267.5s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.53317.5s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.63270.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.13281.0s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.14266.3s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MistralLarge 4 (max + best tools)

Medical procedures markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -50.2%
    Average savings (when looking for best prices): -50.2%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 0%

    True Pricesmax

  • Cost per message$1.87194.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -5.3%
    Average savings (when looking for best prices): -19.7%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 14.3%

    MetaMuse Wonder (max + best tools)

  • Cost per message$1.96231.1s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -5.3%
    Average savings (when looking for best prices): 0.4%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 0%

    Grok4.7 (max + best tools)

  • Cost per message$0.42152.1s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -5.3%
    Average savings (when looking for best prices): -19.7%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 14.9%

    GoogleArgon (max + best tools)

  • Cost per message$2.96251.5s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -5.3%
    Average savings (when looking for best prices): 0.4%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 0%

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.74490.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -5.3%
    Average savings (when looking for best prices): 0.4%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 0%

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.98380.1s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -5.3%
    Average savings (when looking for best prices): -7.3%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 60.4%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.17260.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -5.3%
    Average savings (when looking for best prices): 0.4%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 13.4%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.14251.8s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -5.3%
    Average savings (when looking for best prices): 0.4%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 8.9%

    MistralLarge 4 (max + best tools)

Event tickets markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -14.6%
    Average savings (when looking for best prices): -14.6%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Not applicable

    True Pricesmax

  • Cost per message$2.55134.4s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MetaMuse Wonder (max + best tools)

  • Cost per message$0.59143.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    Grok4.7 (max + best tools)

  • Cost per message$0.51119.8s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -5.9%
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 50%

    GoogleArgon (max + best tools)

  • Cost per message$2.48176.4s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.74195.7s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): -5.9%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 0%

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.31190.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.15233.0s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message——
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    MistralLarge 4 (max + best tools)

Tours & activities markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -20.8%
    Average savings (when looking for best prices): -20.8%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 0%

    True Pricesmax

  • Cost per message$0.97108.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): No matching fixed price
    Average savings (when looking for best prices): -9.2%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 83.3%

    MetaMuse Wonder (max + best tools)

  • Cost per message$0.47241.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): No matching fixed price
    Average savings (when looking for best prices): No matching fixed price
    % of time that best price found when actively searching for it: No matching fixed price
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    Grok4.7 (max + best tools)

  • Cost per message$0.49388.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): -9.2%
    Average savings (when looking for best prices): -9.2%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 77.5%

    GoogleArgon (max + best tools)

  • Cost per message$1.67177.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): No matching fixed price
    % of time that best price found when actively searching for it: No matching fixed price
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.17212.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): No matching fixed price
    Average savings (when looking for best prices): No matching fixed price
    % of time that best price found when actively searching for it: No matching fixed price
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.76243.0s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): No matching fixed price
    Average savings (when looking for best prices): No matching fixed price
    % of time that best price found when actively searching for it: No matching fixed price
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.09246.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): No matching fixed price
    Average savings (when looking for best prices): No matching fixed price
    % of time that best price found when actively searching for it: No matching fixed price
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    DeepSeekV4 Pro (max + best tools)

  • Cost per messageNot reported10.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    MistralLarge 4 (max + best tools)

Theme parks markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -3.9%
    Average savings (when looking for best prices): -3.9%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 0%

    True Pricesmax

  • Cost per message$2.36159.0s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MetaMuse Wonder (max + best tools)

  • Cost per message$1.96209.8s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): 0%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 0%

    Grok4.7 (max + best tools)

  • Cost per message$0.62159.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): API unavailable
    % of time that best price found when actively searching for it: API unavailable
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    GoogleArgon (max + best tools)

  • Cost per message$2.19287.1s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    OpenAIAstra Pro (max + best tools)

  • Cost per message$2.38554.1s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.88463.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): API unavailable
    % of time that best price found when actively searching for it: API unavailable
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.18289.4s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.26338.1s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): 0%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 50%

    MistralLarge 4 (max + best tools)

Movie tickets markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -14.3%
    Average savings (when looking for best prices): -14.3%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 0%

    True Pricesmax

  • Cost per message$1.86127.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MetaMuse Wonder (max + best tools)

  • Cost per message$0.4884.8s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    Grok4.7 (max + best tools)

  • Cost per message$0.4399.8s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    GoogleArgon (max + best tools)

  • Cost per message$2.18169.3s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.55196.7s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.90256.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.12314.8s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.15152.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): Model refused to return prices
    % of time that best price found when actively searching for it: Model refused to return prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Model refused to return prices

    MistralLarge 4 (max + best tools)

Water parks markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -59.8%
    Average savings (when looking for best prices): -59.8%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Not applicable

    True Pricesmax

  • Cost per message$1.64152.7s
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MetaMuse Wonder (max + best tools)

  • Cost per message$1.96137.6s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Model refused to return prices
    Average savings (when looking for best prices): -27.9%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 50%

    Grok4.7 (max + best tools)

  • Cost per message$0.48141.4s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): -42%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 75%

    GoogleArgon (max + best tools)

  • Cost per message$2.73279.7s
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.48278.3s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): -42%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 78.6%

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.44300.3s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): -40.2%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 83.3%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.15264.7s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): -52.6%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 80%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.15182.8s
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MistralLarge 4 (max + best tools)

Ski resorts markets

  • Normal prices returned (under normal usage conditions/requesting vs. normal prices)
  • Average savings (when looking for best prices)
  • % of time that best price found when actively searching for it
  • Hallucinations/wrong item/other failures under scrutiny (when results are actually returned)
  • Normal prices returned (under normal usage conditions/requesting vs. normal prices): -14.4%
    Average savings (when looking for best prices): -14.4%
    % of time that best price found when actively searching for it: 100%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): Not applicable

    True Pricesmax

  • Cost per message$2.64147.3s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): 0%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 75%

    MetaMuse Wonder (max + best tools)

  • Cost per message$1.96182.1s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): 0%
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 50%

    Grok4.7 (max + best tools)

  • Cost per message$0.43179.3s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): 0%
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 50%

    GoogleArgon (max + best tools)

  • Cost per message$2.37317.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): 0%
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 50%

    OpenAIAstra Pro (max + best tools)

  • Cost per message$1.59283.9s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): Only incorrect prices
    Average savings (when looking for best prices): 0%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 50%

    AnthropicOpus 5.5 (max + best tools)

  • Cost per message$0.52230.2s
    Normal prices returned (under normal usage conditions/requesting vs. normal prices): 0%
    Average savings (when looking for best prices): 0%
    % of time that best price found when actively searching for it: 0%
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 25%

    PerplexitySonar Deep Research (max + best tools)

  • Cost per message$0.10265.6s
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    DeepSeekV4 Pro (max + best tools)

  • Cost per message$0.11109.3s
    Average savings (when looking for best prices): Only incorrect prices
    % of time that best price found when actively searching for it: Only incorrect prices
    Hallucinations/wrong item/other failures under scrutiny (when results are actually returned): 100%

    MistralLarge 4 (max + best tools)

Bills & utilities markets

  • True Prices maxAt least $300/yearGuaranteed
  • Meta Muse Wonder (max + best tools)Highly variable
  • Grok 4.7 (max + best tools)Highly variable
  • Google Argon (max + best tools)Highly variable
  • OpenAI Astra Pro (max + best tools)Highly variable
  • Anthropic Opus 5.5 (max + best tools)Highly variable
  • Perplexity Sonar Deep Research (max + best tools)Highly variable
  • DeepSeek V4 Pro (max + best tools)Highly variable
  • Mistral Large 4 (max + best tools)Highly variable

Cruises markets

Coming soon

Airport transfers markets

Coming soon

True Prices does not train on your data.

ProviderMay use your data in trainingCan force your data into trainingEvidence
GoogleDangerDanger

Google

Google Gemini help page: when you give Gemini a task it may use a remote browser or computer, and this is only available when Keep Activity is on.
Google Gemini privacy page: with Keep Activity on, your chats and what you share are saved, and Google uses that activity to develop its services, including training generative AI models, with human reviewers.
xAIDangerDanger
MetaDangerSafe

Meta

Meta Muse Code help page: subscriptions default to Contributor mode, which lets Meta use your prompts, code, files and responses to train future models until you turn it off.
Meta Model API terms: Contributor Services offer no way to exclude specific traffic from training.
Meta Muse page: your conversations and agent activity are used to train new models by default, after removing key personal information, unless you opt out in settings.
OpenAIDangerDanger

OpenAI

OpenAI help page: when another person's agent uses a page you shared, their training settings apply, and your own setting does not control theirs.
OpenAI help page: for personal ChatGPT plans, the “Improve the model for everyone” setting decides whether agent conversations, actions and connected-app data may be used to train models.
AnthropicDangerDanger

Anthropic

Anthropic help page: consumer Claude chats and coding sessions are used to improve models if you allow it, or if they are flagged for safety review.
DeepSeekDangerSafe

DeepSeek

DeepSeek privacy policy: depending on where you live, you may have certain rights over your personal data.
DeepSeek privacy policy: those rights include opting out of your personal data being used to train its models.
PerplexityDangerSafe

Perplexity

Perplexity help page: AI data retention is on by default for Free, Pro and Max users, and collects your data for AI training until you switch it off.
MistralDangerDanger

Mistral

Mistral terms: Mistral may use data and outputs from Labs or Preview models for training, and opt-outs chosen for other Mistral products do not apply to them.
True PricesSafeSafeOur commitment