{
  "name": "AI Signal LLM API pricing dataset: recorded price changes",
  "publisher": "AI Signal",
  "homepage": "https://aisignalhq.com/data/",
  "generatedAt": "2026-10-01T10:32:02.584Z",
  "license": "Creative Commons Attribution 4.0 International (CC BY 4.0)",
  "licenseUrl": "https://creativecommons.org/licenses/by/4.0/",
  "attribution": "LLM API pricing data by AI Signal (https://aisignalhq.com/api-pricing/), licensed under CC BY 4.0.",
  "rowCount": 17,
  "rows": [
    {
      "id": "google-flash-promo-ends",
      "effectiveDate": "2027-01-01",
      "provider": "google",
      "kind": "provider",
      "direction": "increase",
      "multiplier": 2,
      "affects": [
        "google/gemini-3.8-flash",
        "google/gemini-3.7-flash",
        "google/gemini-3.6-flash"
      ],
      "figures": [
        {
          "model": "google/gemini-3.8-flash",
          "before": {
            "input": 0.75,
            "output": 3.75
          },
          "after": {
            "input": 1.5,
            "output": 7.5
          }
        },
        {
          "model": "google/gemini-3.7-flash",
          "before": {
            "input": 0.75,
            "output": 3.75
          },
          "after": {
            "input": 1.5,
            "output": 7.5
          }
        },
        {
          "model": "google/gemini-3.6-flash",
          "before": {
            "input": 0.75,
            "output": 3.75
          },
          "after": {
            "input": 1.5,
            "output": 7.5
          }
        }
      ],
      "summary": "Gemini 3.6, 3.7 and 3.8 Flash pricing is promotional until December 31, 2026. On January 1, 2027 it doubles: input from $0.75 to $1.50 and output from $3.75 to $7.50 per million tokens, the list prices Google shows next to the promotional ones.",
      "sourceUrl": "https://ai.google.dev/gemini-api/docs/pricing",
      "pageUrl": "https://aisignalhq.com/price-changes/#google-flash-promo-ends"
    },
    {
      "id": "anthropic-haiku-4-5-release-date-correction",
      "effectiveDate": "2026-10-01",
      "provider": "anthropic",
      "kind": "correction",
      "direction": "none",
      "multiplier": null,
      "affects": [
        "anthropic/claude-haiku-4-5"
      ],
      "figures": [],
      "summary": "Correction of this site's record, not a price change by Anthropic: no price moved. This page gave Claude Haiku 4.5's release date as October 1, 2025; Anthropic's model page, read on October 1, 2026, prints 'Released October 15, 2025', which also matches the retirement commitment on the same page ('Not sooner than October 15, 2026'). The earlier date had no source recorded. The release date shown here was corrected that day.",
      "sourceUrl": "https://platform.claude.com/docs/en/models/haiku-4-5/overview",
      "pageUrl": "https://aisignalhq.com/price-changes/#anthropic-haiku-4-5-release-date-correction"
    },
    {
      "id": "anthropic-opus-4-5-release-date-correction",
      "effectiveDate": "2026-09-30",
      "provider": "anthropic",
      "kind": "correction",
      "direction": "none",
      "multiplier": null,
      "affects": [
        "anthropic/claude-opus-4-5"
      ],
      "figures": [],
      "summary": "Correction of this site's record, not a price change by Anthropic: no price moved. This page gave Claude Opus 4.5's release date as November 1, 2025; Anthropic's model page, read on September 30, 2026, prints 'Released November 24, 2025', which also matches the retirement commitment on the same page ('Not sooner than November 24, 2026'). The earlier date had no source recorded. The release date shown here was corrected that day.",
      "sourceUrl": "https://platform.claude.com/docs/en/models/opus-4-5/overview",
      "pageUrl": "https://aisignalhq.com/price-changes/#anthropic-opus-4-5-release-date-correction"
    },
    {
      "id": "deepinfra-deepseek-v4-flash-0731-input-down",
      "effectiveDate": "2026-09-30",
      "provider": "deepinfra",
      "kind": "provider",
      "direction": "decrease",
      "multiplier": 1.3333,
      "affects": [
        "deepseek-ai/DeepSeek-V4-Flash-0731"
      ],
      "figures": [
        {
          "model": "deepseek-ai/DeepSeek-V4-Flash-0731",
          "before": {
            "input": 0.08,
            "output": 0.18
          },
          "after": {
            "input": 0.06,
            "output": 0.18
          }
        }
      ],
      "summary": "DeepInfra lowered the input price of DeepSeek V4 Flash 0731 from $0.08 to $0.06 per million tokens; output stays at $0.18. It was $0.08 when this site read DeepInfra's pricing page on September 6, 2026 and $0.06 when it read it again on September 30, 2026; the exact day of the change is not known, so the date here is the day it was seen.",
      "sourceUrl": "https://deepinfra.com/pricing",
      "pageUrl": "https://aisignalhq.com/price-changes/#deepinfra-deepseek-v4-flash-0731-input-down"
    },
    {
      "id": "google-gemini-omni-flash-out-of-scope",
      "effectiveDate": "2026-09-30",
      "provider": "google",
      "kind": "criterion",
      "direction": "withdrawn",
      "multiplier": null,
      "affects": [
        "google/gemini-omni-1.1-flash",
        "google/gemini-omni-flash-preview"
      ],
      "figures": [
        {
          "model": "google/gemini-omni-1.1-flash",
          "before": {
            "input": 1.5,
            "output": 9
          },
          "after": null
        },
        {
          "model": "google/gemini-omni-flash-preview",
          "before": {
            "input": 1.5,
            "output": 9
          },
          "after": null
        }
      ],
      "summary": "Change of this site's criterion, not a price change by Google. Gemini Omni Flash (gemini-omni-1.1-flash) and its preview endpoint (gemini-omni-flash-preview) entered this dataset in the hand sweep of September 16, 2026, priced $1.50 per million input tokens and $9.00 per million output tokens. On September 30, 2026 Google's model page for both versions was read: it describes a model 'for fast, conversational video generation and editing' and gives its supported output as 'Video' only. The pricing table covers models charged per token with text output, so both rows left the table, the calculator and the counts that day. The $9.00 (text) figure on Google's pricing page is labelled as including thinking tokens; it is not the price of a text answer comparable with the other rows. Both rows stay in the dataset with their prices and dates, and their pages remain online, out of the search index.",
      "sourceUrl": "https://ai.google.dev/gemini-api/docs/models/gemini-omni-flash",
      "pageUrl": "https://aisignalhq.com/price-changes/#google-gemini-omni-flash-out-of-scope"
    },
    {
      "id": "alibaba-qwen3.7-flash-added",
      "effectiveDate": "2026-09-17",
      "provider": "alibaba",
      "kind": "omission",
      "direction": "none",
      "multiplier": null,
      "affects": [
        "alibaba/qwen3.7-flash"
      ],
      "figures": [
        {
          "model": "alibaba/qwen3.7-flash",
          "before": null,
          "after": {
            "input": 0.03,
            "output": 0.13
          }
        }
      ],
      "summary": "Omission on our side, not a price change by Alibaba. Qwen3.7 Flash belongs to the same generation as the Qwen3.7 Plus and Max rows and had no row in this dataset. It turned up on September 17, 2026 through the OpenRouter signal of this site's read-only scan for missing models, and its prices were read that day on Alibaba's own Model Studio page: $0.03 input, $0.13 output and $0.006 cached per million tokens in the Singapore (International) region for prompts up to 32K tokens, $0.10 / $0.40 above 32K and $0.20 / $0.80 above 256K. For Alibaba the table covers the current Qwen generations, 3.7 and 3.8; the older generations Alibaba still sells, found the same day, were left out on purpose.",
      "sourceUrl": "https://www.alibabacloud.com/help/en/model-studio/qwen3-7-flash",
      "pageUrl": "https://aisignalhq.com/price-changes/#alibaba-qwen3.7-flash-added"
    },
    {
      "id": "meta-muse-spark-price-found",
      "effectiveDate": "2026-09-16",
      "provider": "meta",
      "kind": "correction",
      "direction": "none",
      "multiplier": null,
      "affects": [
        "meta/muse-spark-1.3",
        "meta/muse-spark-1.2"
      ],
      "figures": [
        {
          "model": "meta/muse-spark-1.3",
          "before": null,
          "after": {
            "input": 1.25,
            "output": 4.25
          }
        },
        {
          "model": "meta/muse-spark-1.2",
          "before": null,
          "after": {
            "input": 1.25,
            "output": 4.25
          }
        }
      ],
      "summary": "Correction of our own reading, not a price change by Meta. From September 11 to September 16, 2026 this site showed Muse Spark 1.3 and Muse Spark 1.2 as 'no published price', because Meta's model pages serve their 'Models and pricing' table only to a browser, and the HTTP client this site verifies with received a JavaScript shell with no pricing text. Read by hand in a browser on September 16, 2026, each page lists two SKUs per model: 'Not used to improve our products' at $1.25 input, $0.15 cached input and $4.25 output per million tokens, and a 'contributor' SKU, 'Used to improve our products', at $0.10 / $0.002 / $0.20, conditional on letting Meta use the traffic to improve its products. The rows now carry the first as the list price, with the contributor price quoted in their notes; Meta changed nothing. Muse Spark 1.1's own model page, read the same day in a browser, carries a benchmark table and no pricing table, and neither of the other two tables lists it, so that row still shows no published price.",
      "sourceUrl": "https://developer.meta.com/ai/models/muse-spark/",
      "pageUrl": "https://aisignalhq.com/price-changes/#meta-muse-spark-price-found"
    },
    {
      "id": "sweep-2026-09-16-missing-models",
      "effectiveDate": "2026-09-16",
      "provider": "multiple",
      "kind": "omission",
      "direction": "none",
      "multiplier": null,
      "affects": [
        "openai/gpt-6-astra",
        "openai/gpt-rosalind-research",
        "anthropic/claude-haiku-3-5",
        "google/gemini-2.5-computer-use-preview",
        "google/gemini-robotics-er-2-preview",
        "google/gemini-robotics-er-2-streaming-preview",
        "google/gemini-omni-flash-preview",
        "mistral/glm-5.3"
      ],
      "figures": [],
      "summary": "Omission on our side, not a price change by any provider. On September 16, 2026, while reading OpenAI's pricing page by hand for a check on batch and flex prices, GPT-6 Astra turned up in the Standard table with no row in this dataset. A hand sweep of the first-party pricing pages of OpenAI, Anthropic, Google, xAI, Mistral, DeepSeek, Alibaba and Cohere against the dataset then found eight models charged per token with text output and a published price that had no row: GPT-6 Astra and GPT-Rosalind (OpenAI), Claude Haiku 3.5 (Anthropic, retired on the first-party API but still priced), Gemini 2.5 Computer Use Preview, Gemini Robotics ER 2 Preview and its Streaming variant, and Gemini Omni Flash Preview, a separate endpoint id from the generally available Omni Flash row at the same prices (Google), and GLM 5.3 (served by Mistral). Astra had been in OpenAI's Standard table at least since September 10, 2026, when the weekly detector first received that page's text, and this site's OpenRouter explainer already named it from the September 15 snapshot. The weekly detector could not have caught any of these: it compares the rows that exist against the page and has no step for models the page lists and the dataset does not. Each new row carries the page quote it was read from. Audio, image, video, embedding and per-minute models found in the same sweep were left out on purpose: the table covers models charged per token with text output.",
      "sourceUrl": "https://developers.openai.com/api/docs/pricing",
      "pageUrl": "https://aisignalhq.com/price-changes/#sweep-2026-09-16-missing-models"
    },
    {
      "id": "openrouter-deepseek-v4-pro-tariff-rule",
      "effectiveDate": "2026-09-14",
      "provider": "openrouter",
      "kind": "criterion",
      "direction": "increase",
      "multiplier": 2,
      "affects": [
        "deepseek/deepseek-v4-pro-0813"
      ],
      "figures": [
        {
          "model": "deepseek/deepseek-v4-pro-0813",
          "before": {
            "input": 0.66,
            "output": 1.98
          },
          "after": {
            "input": 1.32,
            "output": 3.96
          }
        }
      ],
      "summary": "The figure this site publishes for DeepSeek V4 Pro 0813 on OpenRouter changes from $0.66 / $1.98 to $1.32 / $3.96 per million tokens. This is a correction of our own reading, not a price change by DeepSeek or by any host. DeepSeek's own pricing page defines $1.32 / $3.96 as the peak rate and $0.66 / $1.98 as the off-peak rate, half of it, applying outside Monday to Friday 01:00-04:00 and 06:00-10:00 UTC. Every reading behind the old figure had been taken off-peak. Read-only probes on September 11 and 14, 2026 (02:30, 07:30, 12:30 UTC) showed that DeepSeek's endpoint on OpenRouter follows that clock: $1.32 / $3.96 during peak hours, $0.66 / $1.98 the rest of the time. Since this site publishes the maker's list price without discounts, the row now shows the peak rate, which is also what Together AI and Fireworks AI charge at all hours; the off-peak rate is recorded on the row and shown as a note. The peak rate applies 35 of the 168 hours in a week.",
      "sourceUrl": "https://api-docs.deepseek.com/quick_start/pricing/",
      "pageUrl": "https://aisignalhq.com/price-changes/#openrouter-deepseek-v4-pro-tariff-rule"
    },
    {
      "id": "alibaba-reference-region-singapore",
      "effectiveDate": "2026-09-11",
      "provider": "alibaba",
      "kind": "criterion",
      "direction": "increase",
      "multiplier": null,
      "affects": [
        "alibaba/qwen3.8-max",
        "alibaba/qwen3.8-flash",
        "alibaba/qwen3.7-plus",
        "alibaba/qwen3.7-max"
      ],
      "figures": [
        {
          "model": "alibaba/qwen3.8-max",
          "before": {
            "input": 1.65,
            "output": 4.951
          },
          "after": {
            "input": 2,
            "output": 6
          }
        },
        {
          "model": "alibaba/qwen3.8-flash",
          "before": {
            "input": 0.113,
            "output": 0.382
          },
          "after": {
            "input": 0.15,
            "output": 0.47
          }
        },
        {
          "model": "alibaba/qwen3.7-plus",
          "before": {
            "input": 0.276,
            "output": 1.101
          },
          "after": {
            "input": 0.4,
            "output": 1.6
          }
        },
        {
          "model": "alibaba/qwen3.7-max",
          "before": {
            "input": 1.65,
            "output": 4.951
          },
          "after": {
            "input": 2.5,
            "output": 7.5
          }
        }
      ],
      "summary": "Change of criterion on our side, not a price change by Alibaba: no Alibaba price rose. Alibaba Cloud's Model Studio pages list the same model at two price levels: the China (Beijing) region and the Global regions (Frankfurt, Virginia, Tokyo, Hong Kong) share one level, and the Singapore region, labelled by Alibaba \"Scope: International\", is higher. Until September 11, 2026 the first-party table quoted the Beijing/Global level while the provider comparison quoted Alibaba's OpenRouter endpoint at the Singapore level, so the site showed two different \"Alibaba\" prices for the same model. From September 11 the reference region is Singapore (International), the one Alibaba labels international, the one every reseller we track (OpenRouter, Together, Fireworks) mirrors, and the higher of the two; the Beijing/Global figures stay published next to it. Published figures move from 1.65 / 4.951 to 2.00 / 6.00 (Qwen3.8 Max), 0.113 / 0.382 to 0.15 / 0.47 (Qwen3.8 Flash), 0.276 / 1.101 to 0.40 / 1.60 (Qwen3.7 Plus, prompts up to 256K; the tier above 256K, 1.20 / 4.80, is now recorded too) and 1.65 / 4.951 to 2.50 / 7.50 (Qwen3.7 Max). Alibaba's pages state that they show the original pricing excluding any limited-time promotions.",
      "sourceUrl": "https://www.alibabacloud.com/help/en/model-studio/qwen3-8-max",
      "pageUrl": "https://aisignalhq.com/price-changes/#alibaba-reference-region-singapore"
    },
    {
      "id": "openrouter-deepseek-v4-flash-maker-endpoint-withdrawn",
      "effectiveDate": "2026-09-11",
      "provider": "openrouter",
      "kind": "provider",
      "direction": "withdrawn",
      "multiplier": null,
      "affects": [
        "deepseek/deepseek-v4-flash-0731"
      ],
      "figures": [
        {
          "model": "deepseek/deepseek-v4-flash-0731",
          "before": {
            "input": 0.22,
            "output": 0.66
          },
          "after": null
        }
      ],
      "summary": "DeepSeek's own endpoint for DeepSeek V4 Flash 0731 disappeared from OpenRouter between the afternoon of September 9, 2026, when it was still listed at $0.22 / $0.66 and this row was verified against it, and the 04:30 UTC check of September 10, which found no DeepSeek endpoint; it had not returned by September 11. DeepSeek still serves V4 Pro 0813 on OpenRouter, so this is specific to this model, not a general withdrawal. This site's OpenRouter column shows the model maker's list price, and for this model there is now none to show: the column moves from $0.22 / $0.66 per million tokens (last verified September 9) to \"no list price\". Nothing changed in our criterion, and no host changed a price; the provider withdrew its endpoint. The model is still served on OpenRouter by resellers (28 listed, 26 healthy on September 11), reachable with :floor; that day the catalog quoted $0.065 / $0.18 via Relace.",
      "sourceUrl": "https://openrouter.ai/api/v1/models/deepseek/deepseek-v4-flash-0731/endpoints",
      "pageUrl": "https://aisignalhq.com/price-changes/#openrouter-deepseek-v4-flash-maker-endpoint-withdrawn"
    },
    {
      "id": "openai-gpt56-promo-ends",
      "effectiveDate": "2026-09-10",
      "provider": "openai",
      "kind": "correction",
      "direction": "increase",
      "multiplier": 2,
      "affects": [
        "openai/gpt-5.6-sol",
        "openai/gpt-5.6-terra",
        "openai/gpt-5.6-luna"
      ],
      "figures": [
        {
          "model": "openai/gpt-5.6-sol",
          "before": {
            "input": 2,
            "output": 10
          },
          "after": {
            "input": 4,
            "output": 20
          }
        },
        {
          "model": "openai/gpt-5.6-terra",
          "before": {
            "input": 1,
            "output": 6
          },
          "after": {
            "input": 2,
            "output": 12
          }
        },
        {
          "model": "openai/gpt-5.6-luna",
          "before": {
            "input": 0.1,
            "output": 0.6
          },
          "after": {
            "input": 0.2,
            "output": 1.2
          }
        }
      ],
      "summary": "Correction of our own reading, not a price change by OpenAI. From September 3 to September 10, 2026 this site published GPT-5.6 Sol at $2 / $10, Terra at $1 / $6 and Luna at $0.10 / $0.60 per million tokens, read from the Batch tab of OpenAI's pricing page, which is always 50% of Standard (the page's four tabs, Standard, Batch, Flex and Fast mode, look identical in plain text), and labelled as a promotion ending November 21 with double the price after it. The Standard short-context prices, verified by hand on September 10, are Sol $4 / $20, Terra $2 / $12 and Luna $0.20 / $1.20, and the published figures doubled to them. The November 21 date was OpenAI's, but it belongs to Sol alone and is not an end date: OpenAI states that Sol's $4 / $20 is promotional pricing, a 20% reduction on input and 33% on output, available at least through November 21, 2026, and publishes no list price for after it; Terra and Luna carry no promotion. On September 10 this entry wrongly said there was no promotion to end; corrected on September 15, when the Sol row got its promotion back, marked as at least through that date, without an invented list price.",
      "sourceUrl": "https://developers.openai.com/api/docs/pricing",
      "pageUrl": "https://aisignalhq.com/price-changes/#openai-gpt56-promo-ends"
    },
    {
      "id": "openrouter-deepseek-v4-flash-list-price-rule",
      "effectiveDate": "2026-09-09",
      "provider": "openrouter",
      "kind": "criterion",
      "direction": "increase",
      "multiplier": 4.4,
      "affects": [
        "deepseek/deepseek-v4-flash-0731"
      ],
      "figures": [
        {
          "model": "deepseek/deepseek-v4-flash-0731",
          "before": {
            "input": 0.05,
            "output": 0.16
          },
          "after": {
            "input": 0.22,
            "output": 0.66
          }
        }
      ],
      "summary": "The published OpenRouter price of DeepSeek V4 Flash 0731 rises 4.4x on input ($0.05 to $0.22) and 4.1x on output ($0.16 to $0.66). No host raised its rate. The number changes because the rule changed: until September 9, 2026 this column showed the cheapest healthy endpoint at the moment of checking (OpenInference); from September 9 it shows the model maker's own price on OpenRouter's providers page, without promotional discounts (DeepSeek, $0.22 / $0.66). Cheaper resellers still exist and are still reachable with :floor.",
      "sourceUrl": "https://openrouter.ai/api/v1/models/deepseek/deepseek-v4-flash-0731/endpoints",
      "pageUrl": "https://aisignalhq.com/price-changes/#openrouter-deepseek-v4-flash-list-price-rule"
    },
    {
      "id": "openrouter-glm-5.3-flash-list-price-rule",
      "effectiveDate": "2026-09-09",
      "provider": "openrouter",
      "kind": "criterion",
      "direction": "increase",
      "multiplier": 2.1,
      "affects": [
        "z-ai/glm-5.3-flash"
      ],
      "figures": [
        {
          "model": "z-ai/glm-5.3-flash",
          "before": {
            "input": 0.0713,
            "output": 0.2375
          },
          "after": {
            "input": 0.15,
            "output": 0.5
          }
        }
      ],
      "summary": "The published OpenRouter price of GLM-5.3 Flash rises 2.1x on both input ($0.0713 to $0.15) and output ($0.2375 to $0.50). No host raised its rate. The number changes because the rule changed: until September 9, 2026 this column showed the cheapest healthy endpoint at the moment of checking (Relace); from September 9 it shows the model maker's list price on OpenRouter's providers page, without promotional discounts (Z.ai, $0.15 / $0.50). On September 9 Z.ai was still showing a 50% launch discount ($0.075 / $0.25, struck-through list price) that it had announced would end at midnight Singapore time, 16:00 UTC. Cheaper resellers still exist and are still reachable with :floor.",
      "sourceUrl": "https://openrouter.ai/api/v1/models/z-ai/glm-5.3-flash/endpoints",
      "pageUrl": "https://aisignalhq.com/price-changes/#openrouter-glm-5.3-flash-list-price-rule"
    },
    {
      "id": "openrouter-glm-5.3-list-price-rule",
      "effectiveDate": "2026-09-09",
      "provider": "openrouter",
      "kind": "criterion",
      "direction": "increase",
      "multiplier": 1.25,
      "affects": [
        "z-ai/glm-5.3"
      ],
      "figures": [
        {
          "model": "z-ai/glm-5.3",
          "before": {
            "input": 1.12,
            "output": 3.52
          },
          "after": {
            "input": 1.4,
            "output": 4.4
          }
        }
      ],
      "summary": "The published OpenRouter price of GLM-5.3 rises 1.25x on both input ($1.12 to $1.40) and output ($3.52 to $4.40). No host raised its rate. The number changes because the rule changed: until September 9, 2026 this column showed the cheapest healthy endpoint at the moment of checking (GMICloud); from September 9 it shows the model maker's own price on OpenRouter's providers page, without promotional discounts (Z.ai, $1.40 / $4.40). Cheaper resellers still exist and are still reachable with :floor.",
      "sourceUrl": "https://openrouter.ai/api/v1/models/z-ai/glm-5.3/endpoints",
      "pageUrl": "https://aisignalhq.com/price-changes/#openrouter-glm-5.3-list-price-rule"
    },
    {
      "id": "openrouter-deepseek-v4-pro-cheapest-host-changes",
      "effectiveDate": "2026-09-08",
      "provider": "openrouter",
      "kind": "criterion",
      "direction": "increase",
      "multiplier": 1.136,
      "affects": [
        "deepseek/deepseek-v4-pro-0813"
      ],
      "figures": [
        {
          "model": "deepseek/deepseek-v4-pro-0813",
          "before": {
            "input": 0.5808,
            "output": 1.7424
          },
          "after": {
            "input": 0.66,
            "output": 1.98
          }
        }
      ],
      "summary": "The figure this site published for DeepSeek V4 Pro 0813 on OpenRouter rose 13.6% on September 8, 2026, from $0.5808 / $1.7424 to $0.66 / $1.98 per million tokens. DeepSeek did not change any price. At the time the published figure was the cheapest healthy OpenRouter endpoint, and that endpoint switched from Alibaba to DeepSeek's own: DeepSeek's endpoint read $0.66 / $1.98 on both days, while Alibaba's went from $0.5808 / $1.7424 to $1.122 / $3.366 and has since been seen at both figures again. This entry originally described Alibaba's move as a 12% discount ending; that reading was wrong. Both readings were also taken outside DeepSeek's peak hours, so $0.66 / $1.98 was the maker's off-peak rate, not its list price. This rule was replaced on September 9 (maker's list price) and again on September 14 (see openrouter-deepseek-v4-pro-tariff-rule). Kept because the number on this site did change that day; it was our rule, not a host's price.",
      "sourceUrl": "https://openrouter.ai/api/v1/models/deepseek/deepseek-v4-pro-0813/endpoints",
      "pageUrl": "https://aisignalhq.com/price-changes/#openrouter-deepseek-v4-pro-cheapest-host-changes"
    },
    {
      "id": "fireworks-us-only-serverless-surcharge",
      "effectiveDate": "2026-09-01",
      "provider": "fireworks",
      "kind": "provider",
      "direction": "increase",
      "multiplier": 1.5,
      "affects": [],
      "figures": [],
      "summary": "Since September 1, 2026 Fireworks adds a 50% surcharge to the base price of the Serverless models it marks as US-only. The Fireworks prices on this site are the base prices, without the surcharge.",
      "sourceUrl": "https://docs.fireworks.ai/serverless/pricing",
      "pageUrl": "https://aisignalhq.com/price-changes/#fireworks-us-only-serverless-surcharge"
    }
  ]
}