Where the prices come from, and how fresh they are
Every figure on the site is a public list price, collected automatically from each cloud's own price list or documentation every three hours. No negotiated rates, commitments or taxes.
73,749 prices are held now, across 24 vendors and 143 regions: 67,973 per token, 4,153 reserved capacity, 1,085 data transfer, 538 gpu instances. 2,191,291 are kept in history, from 715 collections.
Prices held, by cloud
| Cloud | Prices | Regions | Last collected | Age | Status |
|---|---|---|---|---|---|
| Alibaba Cloud | 210 | 12 | 21 November 2026 | 0 hours ago | live |
| Anthropic | 124 | 1 | 21 November 2026 | 0 hours ago | live |
| AWS | 25,006 | 36 | 21 November 2026 | 0 hours ago | live |
| Azure | 45,607 | 50 | 21 November 2026 | 0 hours ago | live |
| Cloudflare | 90 | 1 | 21 November 2026 | 0 hours ago | live |
| Core42 | 110 | 4 | 21 November 2026 | 0 hours ago | live |
| Crusoe | 42 | 1 | 21 November 2026 | 0 hours ago | live |
| Databricks | 279 | 1 | 21 November 2026 | 0 hours ago | live |
| DeepInfra | 707 | 1 | 21 November 2026 | 0 hours ago | live |
| Fireworks AI | 198 | 2 | 21 November 2026 | 0 hours ago | live |
| Google Cloud | 254 | 2 | 21 November 2026 | 0 hours ago | live |
| Gemini API | 156 | 1 | 21 November 2026 | 0 hours ago | live |
| Groq | 24 | 1 | 21 November 2026 | 0 hours ago | live |
| HF Endpoints | 28 | 1 | 21 November 2026 | 0 hours ago | live |
| Huawei Cloud | 21 | 1 | 21 November 2026 | 0 hours ago | live |
| Mistral | 24 | 1 | 21 November 2026 | 0 hours ago | live |
| Nebius | 91 | 9 | 21 November 2026 | 0 hours ago | live |
| OCI | 137 | 11 | 21 November 2026 | 0 hours ago | live |
| OpenAI | 156 | 1 | 21 November 2026 | 0 hours ago | live |
| Snowflake | 311 | 1 | 21 November 2026 | 0 hours ago | live |
| Telnyx | 54 | 2 | 21 November 2026 | 0 hours ago | live |
| Together AI | 70 | 1 | 21 November 2026 | 0 hours ago | live |
| IBM watsonx | 26 | 1 | 21 November 2026 | 0 hours ago | live |
| xAI | 24 | 1 | 21 November 2026 | 0 hours ago | live |
News data
2,425 changes are on record (143 new regions, 45 instances added, 692 models added, 282 price cuts, 161 price rises, 210 retirements, 498 announcements, 394 other), 537 retirement dates are tracked, 114 of them within 90 days, and 1,362 announcements are kept by headline, date and link. They are read with every price refresh.
| Source | Retirement dates | Announcements | Status |
|---|---|---|---|
| OpenAI | 90 | 243 | live |
| Anthropic | 19 | 221 | live |
| Azure | 191 | 38 | live |
| Alibaba | 0 | 188 | live |
| Together AI | 0 | 137 | live |
| 24 | 106 | live | |
| Databricks | 0 | 120 | live |
| AWS | 47 | 57 | live |
| Snowflake | 0 | 68 | live |
| OCI | 35 | 29 | live |
| DeepInfra | 58 | live | |
| Mistral | 15 | 42 | live |
| Cloudflare | 0 | 38 | live |
| Fireworks AI | 0 | 25 | live |
| Telnyx | 0 | 24 | live |
| Gemini API | 22 | live | |
| Nebius | 21 | live | |
| xAI | 0 | 19 | live |
| IBM watsonx | 10 | live | |
| Huawei Cloud | 0 | 7 | live |
| Groq | 5 | live |
The public sources
AWS
- GPU instances: https://pricing.us-east-1.amazonaws.com/offers/v1.0/aws/AmazonEC2/current/{region}/index.csv
- Bedrock token prices: https://pricing.us-east-1.amazonaws.com/offers/v1.0/aws/{AmazonBedrock,AmazonBedrockService,AmazonBedrockFoundationModels}/current/index.json
- freshness overlay: https://b0.p.awsstatic.com/pricing/2.0/meteredUnitMaps/bedrock/USD/current/bedrock.json
Azure
- GPU VMs and token meters: https://prices.azure.com/api/retail/prices
OCI
- GPU shapes and Generative AI meters: https://apexapps.oracle.com/pls/apex/cetools/api/v1/products/
- models offered, and in which mode: https://docs.oracle.com/en-us/iaas/Content/generative-ai/pretrained-models.htm
- on-demand vs dedicated, per region: https://docs.oracle.com/en-us/iaas/Content/generative-ai/model-endpoint-regions.htm
- dedicated cluster AI unit counts: https://docs.oracle.com/en-us/iaas/Content/generative-ai/hardware-unit-shapes-by-region.htm
Core42
- per-token prices: https://www.core42.ai/compass/documentation/compass-model-pricing
- models, tasks and the region each runs in: https://www.core42.ai/compass/documentation/compass-models
- deprecation and retirement dates: https://www.core42.ai/compass/documentation/compass-model-deprecations-and-retirements
- GPUs by the hour (AI Cloud): https://www.core42.ai/products/ai-cloud
Alibaba Cloud
- per-token prices (Model Studio): https://www.alibabacloud.com/help/en/model-studio/model-pricing
- GPU instance prices (ECS OpenAPI): https://www.alibabacloud.com/help/en/ecs/developer-reference/api-ecs-2014-05-26-describeprice
- GPU instance specifications: https://www.alibabacloud.com/help/en/ecs/user-guide/gpu-accelerated-compute-optimized-and-vgpu-accelerated-instance-families-1
Google Cloud
- per-token prices (Vertex AI): https://cloud.google.com/vertex-ai/generative-ai/pricing
- GPU machine prices (Compute Engine): https://cloud.google.com/products/compute/pricing/accelerator-optimized
- GPU machine specifications: https://cloud.google.com/compute/docs/gpus
- internet egress (Premium Tier): https://cloud.google.com/vpc/network-pricing
Nebius
- per-token prices (Token Factory): https://tokenfactory.nebius.com/api/public/models_info
- GPU virtual machine prices: https://docs.nebius.com/compute/resources/pricing.md
- GPU virtual machine specifications: https://docs.nebius.com/compute/virtual-machines/types
- egress: https://nebius.com/prices
Mistral
- per-token prices, bought from Mistral directly: https://docs.mistral.ai/inference/pricing.md
Telnyx
- per-token prices (Inference API): https://telnyx.com/pricing/inference-api
- model ids: https://developers.telnyx.com/docs/inference/models.md
- models served from Dubai: https://telnyx.com/release-notes/uae-data-locality
Snowflake
- per-token prices (Cortex AI, Table 6 of the consumption table): https://www.snowflake.com/legal-files/CreditConsumptionTable.pdf
- AI Credit price and routing: https://docs.snowflake.com/en/user-guide/snowflake-cortex/pricing
Cloudflare
- per-token prices (Workers AI): https://developers.cloudflare.com/workers-ai/platform/pricing/index.md
Databricks
- per-token prices, open models (DBUs per 1M tokens) and the $ per DBU example: https://www.databricks.com/product/pricing/foundation-model-serving
- per-token prices, proprietary models (OpenAI, Anthropic, Google, xAI): https://www.databricks.com/product/pricing/proprietary-foundation-model-serving
Gemini API
- per-token prices, bought from Google directly (Gemini API): https://ai.google.dev/gemini-api/docs/pricing.md.txt
- retirements: https://ai.google.dev/gemini-api/docs/deprecations.md.txt
HF Endpoints
- dedicated GPU endpoint prices (Inference Endpoints): https://huggingface.co/docs/inference-endpoints/support/pricing.md
IBM watsonx
- per-token prices (supported foundation models): https://dataplatform.cloud.ibm.com/docs/content/wsj/analyze-data/fm-models.html
- deploy-on-demand GPU rates (runtime plans): https://dataplatform.cloud.ibm.com/docs/content/wsj/getting-started/wxai-runtime-plans-genai.html
xAI
- per-token prices, bought from xAI directly: https://docs.x.ai/developers/models.md
Together AI
- per-token prices (serverless models): https://docs.together.ai/docs/serverless/models.md
- GPU cluster prices: https://www.together.ai/pricing
Crusoe
- GPU instance and Managed Inference prices: https://www.crusoe.ai/cloud/pricing
Huawei Cloud
- per-token prices (MaaS): https://support.huaweicloud.com/intl/en-us/price-maas/price-maas-0002.html
Groq
- per-token prices: https://console.groq.com/docs/models
- retirement dates: https://console.groq.com/docs/deprecations
Fireworks AI
- per-token prices (serverless): https://docs.fireworks.ai/serverless/pricing
- GPU prices (on-demand deployments): https://fireworks.ai/pricing
- announcements: https://docs.fireworks.ai/updates/changelog
DeepInfra
- per-token prices and deprecations (model list): https://api.deepinfra.com/models/list
- GPU prices (custom-LLM deployments): https://deepinfra.com/pricing
OpenAI
- per-token prices, bought from OpenAI directly: https://developers.openai.com/api/docs/pricing.md
Anthropic
- per-token prices, bought from Anthropic directly: https://platform.claude.com/docs/en/about-claude/pricing.md