{
  "title": "Jev AI pricing explained: $0.042 per million, free output",
  "canonical_url": "https://productwithattitude.com/articles/jev-pricing-cost-per-token",
  "status": "published",
  "published_at": "2026-09-25",
  "updated_at": "2026-09-25",
  "author": {
    "name": "Karo Zieminski",
    "url": "https://productwithattitude.com/about"
  },
  "summary": "Jev costs $0.042 per million input tokens. See real bills, what AI providers pay for, and why System One models can cost less.",
  "markdown_url": "https://productwithattitude.com/articles/jev-pricing-cost-per-token.md",
  "faq": [
    {
      "question": "Is there a free Jev plan?",
      "answer": "I received a $5 sign-up credit for the test in my Substack post. That was the offer I received at the time. Check your TypeSafe account for the terms available to you. A credit pays for usage until it runs out. It does not remove the usage charge. Free output also leaves you paying for the text Jev reads."
    },
    {
      "question": "Does Jev charge separately for every question?",
      "answer": "TypeSafe lists a price for input tokens, with no separate price per question. You can ask several questions in one request, as I did when scoring an article. Extra questions add text, so they can increase the input charge. Use the token count Jev reports to check the bill. When several questions use the same document, sending them together can avoid paying to send that document repeatedly."
    },
    {
      "question": "Can several questions share the same input?",
      "answer": "Yes. TypeSafe lets you send one set of text with several questions. For example, Jev can judge a message’s category and urgency from the same message. The questions must be answerable on their own. If the second question needs the first answer, wait for that answer before sending the next question. Asking independent questions together can save separate trips to Jev."
    },
    {
      "question": "Can confidence scores eliminate review costs?",
      "answer": "A confidence score shows how sure Jev is about an answer. I would test how well that score matches correct answers before removing human review. TypeSafe recommends testing on your own examples. You choose a cutoff: answers above it can pass through, while answers below it need another check. The right cutoff depends on what happens when Jev is wrong. Record how many answers pass without review and how many of those answers are wrong. Changing the cutoff changes both numbers. Saving on checks may create more work fixing mistakes later."
    },
    {
      "question": "Is Jev’s low price sustainable?",
      "answer": "We know what TypeSafe charges today. We do not know whether that price covers its costs and leaves a profit. The company says it needs time to show that the price can last. I would keep the price easy to change in my cost estimate. I would also check what a higher rate would mean. That helps me plan; it does not mean I expect a rise. First, test whether Jev gives useful answers at the current price."
    }
  ],
  "reviewed_at": "2026-09-24",
  "claims": [
    {
      "text": "As checked on 24 September 2026, TypeSafe AI lists Jev at $0.042 per million input tokens and zero output-token charges.",
      "kind": "vendor_statement",
      "sources": [
        "https://docs.typesafe.ai/models"
      ]
    },
    {
      "text": "$0.042 per million input tokens equals $42 per billion input tokens.",
      "kind": "calculation",
      "sources": [
        "https://docs.typesafe.ai/models"
      ]
    },
    {
      "text": "At the listed rate, a request with 3,000 billed input tokens costs $0.000126.",
      "kind": "calculation",
      "sources": [
        "https://docs.typesafe.ai/models"
      ]
    },
    {
      "text": "At the listed rate, 100,000 requests with 3,000 billed input tokens each cost $12.60 for Jev usage alone.",
      "kind": "calculation",
      "sources": [
        "https://docs.typesafe.ai/models"
      ]
    },
    {
      "text": "Karo Zieminski's 21 September 2026 published example reports 2,994 input tokens for a four-question article-scoring request, equivalent to $0.000125748 at the listed rate.",
      "kind": "reported_usage",
      "sources": [
        "https://docs.typesafe.ai/models",
        "https://karozieminski.substack.com/p/what-is-jev-cost"
      ]
    },
    {
      "text": "An illustrative workload of ten 1,000-token requests per second costs $1.512 per hour and $1,088.64 over thirty continuous days at the listed rate.",
      "kind": "calculation",
      "sources": [
        "https://docs.typesafe.ai/models"
      ]
    },
    {
      "text": "An illustrative 100,000-request workload at 3,000 tokens per request plus 5,000 requests to a backup model at an assumed $0.01 each costs $62.60 before other expenses.",
      "kind": "calculation",
      "sources": [
        "https://docs.typesafe.ai/models",
        "https://aws.amazon.com/blogs/aws-cloud-financial-management/calculating-the-roi-of-ai/"
      ]
    },
    {
      "text": "AI services may pay for rented compute time, owned infrastructure, or another provider’s model service; customer token prices are distinct from these costs.",
      "kind": "explanation",
      "sources": [
        "https://aws.amazon.com/ec2/pricing/on-demand/",
        "https://developer.nvidia.com/blog/llm-inference-benchmarking-how-much-does-your-llm-inference-cost/",
        "https://aws.amazon.com/bedrock/pricing/"
      ]
    },
    {
      "text": "Training and running a trained model incur different costs, including compute, memory, data preparation, storage, operations, and supporting infrastructure.",
      "kind": "explanation",
      "sources": [
        "https://docs.aws.amazon.com/wellarchitected/latest/machine-learning-lens/mlcost02-bp01.html",
        "https://developer.nvidia.com/blog/mastering-llm-techniques-inference-optimization/",
        "https://developer.nvidia.com/blog/?p=118314"
      ]
    },
    {
      "text": "Typical autoregressive language models process input and then generate output through successive dependent steps.",
      "kind": "explanation",
      "sources": [
        "https://developer.nvidia.com/blog/mastering-llm-techniques-inference-optimization/"
      ]
    },
    {
      "text": "TypeSafe documents shared input processing and parallel evaluation of questions, and says Jev produces answer probabilities without autoregressive text generation.",
      "kind": "vendor_statement",
      "sources": [
        "https://typesafe.ai/blog/introducing-system-one-models-and-jev",
        "https://docs.typesafe.ai/patterns/fan-out",
        "https://docs.typesafe.ai/concepts/state",
        "https://docs.typesafe.ai/concepts/system-one",
        "https://docs.typesafe.ai/introduction/machine-learning-primer"
      ]
    },
    {
      "text": "Completing more useful decisions with the same hardware time can reduce the average cost per decision; this is an economic explanation rather than a measurement of TypeSafe’s costs.",
      "kind": "explanation",
      "sources": [
        "https://developer.nvidia.com/blog/llm-inference-benchmarking-how-much-does-your-llm-inference-cost/",
        "https://developer.nvidia.com/blog/mastering-llm-techniques-inference-optimization/"
      ]
    },
    {
      "text": "TypeSafe’s exact operating costs, training cost, and profit margin are not established by its published customer price.",
      "kind": "limitation",
      "sources": [
        "https://typesafe.ai/blog/introducing-system-one-models-and-jev"
      ]
    }
  ],
  "sources": [
    {
      "url": "https://docs.typesafe.ai/models",
      "label": "Current model and pricing documentation"
    },
    {
      "url": "https://docs.typesafe.ai/api",
      "label": "API reference and usage accounting"
    },
    {
      "url": "https://typesafe.ai/blog/introducing-system-one-models-and-jev",
      "label": "Vendor design, performance, and commercial disclosures"
    },
    {
      "url": "https://docs.typesafe.ai/patterns/fan-out",
      "label": "Parallel questions"
    },
    {
      "url": "https://docs.typesafe.ai/concepts/state",
      "label": "Shared state and input modalities"
    },
    {
      "url": "https://docs.typesafe.ai/model-jaggedness/jev-1.13",
      "label": "Documented model limitations"
    },
    {
      "url": "https://docs.typesafe.ai/confidence",
      "label": "Confidence thresholds"
    },
    {
      "url": "https://aws.amazon.com/ec2/pricing/on-demand/",
      "label": "Compute time billing"
    },
    {
      "url": "https://developer.nvidia.com/blog/llm-inference-benchmarking-how-much-does-your-llm-inference-cost/",
      "label": "Server ownership, depreciation, and cost per completed request"
    },
    {
      "url": "https://aws.amazon.com/bedrock/pricing/",
      "label": "Managed model service billing"
    },
    {
      "url": "https://docs.aws.amazon.com/wellarchitected/latest/machine-learning-lens/mlcost02-bp01.html",
      "label": "Data, training, hosting, and maintenance costs"
    },
    {
      "url": "https://developer.nvidia.com/blog/mastering-llm-techniques-inference-optimization/",
      "label": "Input processing, text generation, memory, caching, and batching"
    },
    {
      "url": "https://developer.nvidia.com/blog/?p=118314",
      "label": "Power and cooling constraints"
    },
    {
      "url": "https://aws.amazon.com/blogs/aws-cloud-financial-management/calculating-the-roi-of-ai/",
      "label": "AI expenses beyond direct model usage"
    },
    {
      "url": "https://docs.typesafe.ai/concepts/system-one",
      "label": "System One decision model definition"
    },
    {
      "url": "https://docs.typesafe.ai/introduction/machine-learning-primer",
      "label": "TypeSafe decision-focused training objective"
    },
    {
      "url": "https://karozieminski.substack.com/p/what-is-jev-cost",
      "label": "Karo’s published article-scoring example"
    }
  ]
}