Offerings

Every active offering: a model as served by one host, with its current canonical price. Sort by clicking a column header; filter with the row below. Prices are USD per 1M tokens unless the unit toggle is set to per 1M characters (an approximation using a measured, blended tokens-per-character ratio).

Showing tier-tagged and vendor-direct models. Show all offerings, including the long tail

More columns: + Vendor + Region - Cache + Batch + Tokenizer + Last change + Source All columns Default columns

ModelHostVariantRegionInputOutputBlended ↑?Cache readCache writeContextConfidence?
InklingOpenRouter––1.004.051.760.170–524,288api
Inkling 3 sourcesTogether AI––1.004.051.760.170–524,288api
DeepSeek V4 ProAlibaba Cloudfp8–1.422.831.770.118–1,048,576api
Amazon Nova ProAWS Bedrock–eu1.054.201.840.263–300,000aggregator
o4-mini 3 sourcesOpenAI––1.104.401.930.275–200,000api
o4-mini 2 sourcesAzure AI Foundry––1.104.401.930.275–200,000aggregator
DeepSeek V4 ProAlibaba Cloudfp8us-east-11.563.121.950.130–1,048,576api
DeepSeek V4 Pro 2 sourcesDeepSeek––1.323.961.980.0440free1,048,576official
Claude Haiku 4.5OpenRouter––1.005.002.000.1001.25200,000api
GPT-6.1 Sol 2 sourcesOpenAIflex–1.005.002.000.05001.251,050,000official
DeepSeek V4 ProNovita AIfp8–1.603.202.000.135–1,048,576api
Claude Haiku 4.5Azure AI Foundry–global1.005.002.000.1001.25200,000api
Claude Haiku 4.5 2 sourcesAWS Bedrock–global1.005.002.000.1001.25200,000api
Claude Haiku 4.5Google Vertex AI–global1.005.002.000.1001.25200,000api
Claude Haiku 4.5 4 sourcesAnthropic––1.005.002.000.1001.25200,000official
Claude Haiku 4.5 2 sourcesAzure AI Foundry––1.005.002.000.1001.25200,000aggregator
Qwen3-235B-A22BHyperbolic––2.002.002.00––131,072aggregator
Kimi K2Hyperbolic––2.002.002.00––131,072aggregator
Claude Haiku 4.5Perplexity––1.005.002.000.100–200,000aggregator
Claude Haiku 4.5 2 sourcesGoogle Vertex AI––1.005.002.000.1001.25200,000aggregator
DeepSeek V4 Pro 2 sourcesNovita AI––1.603.202.000.135–1,048,576aggregator
Claude Haiku 4.5 2 sourcesDeepInfra––1.005.002.00––200,000official
GPT-6.1 SolOpenAIbatch–1.005.002.000.05001.251,050,000official
DeepSeek V4 ProVenice––1.653.302.060.330–1,048,576api
DeepSeek V4 ProAtlas Cloudfp4–1.683.382.110.130–1,048,576api
DeepSeek V4 ProBaidufp8–1.693.382.110.140–1,048,576api
o4-miniAzure AI Foundry–eu1.214.842.120.303–200,000aggregator
o4-miniAzure AI Foundry–us1.214.842.120.303–200,000aggregator
DeepSeek V4 ProStreamLake promo −45%fp8–1.743.482.180.145–1,048,576api
DeepSeek V4 ProGMI Cloud promo −45%fp8–1.743.482.180.145–1,048,576api
DeepSeek V4 Pro 2 sourcesAzure AI Foundry––1.743.482.180.145–1,048,576aggregator
DeepSeek V4 Pro 2 sourcesNebius AI Studio––1.753.502.190.140–1,048,576aggregator
DeepSeek V4 ProNebius AI Studiofp8–1.753.502.190.140–1,048,576api
Claude Haiku 4.5 2 sourcesAWS Bedrock–us1.105.502.200.1101.38200,000api
Claude Haiku 4.5Google Vertex AI–us-east51.105.502.200.1101.38200,000api
Claude Haiku 4.5AWS Bedrock–eu-west-11.105.502.200.1101.38200,000api
Claude Haiku 4.5Google Vertex AI–europe1.105.502.200.1101.38200,000api
Claude Haiku 4.5 2 sourcesAWS Bedrock––1.105.502.200.1101.38200,000aggregator
Claude Haiku 4.5AWS Bedrock–apac1.105.502.200.1101.38200,000aggregator
Claude Haiku 4.5AWS Bedrock–au1.105.502.200.1101.38200,000aggregator
Claude Haiku 4.5AWS Bedrock–eu1.105.502.200.1101.38200,000aggregator
Claude Haiku 4.5AWS Bedrock–jp1.105.502.200.1101.38200,000aggregator
DeepSeek V4 ProAzure AI Foundry–us1.913.832.390.160–1,048,576api
Qwen3-Max 3 sourcesDeepInfra––1.206.002.400.240–262,144official
Claude Haiku 4.5AWS Bedrock–us-gov-east-11.206.002.400.1201.50200,000aggregator
Claude Haiku 4.5AWS Bedrock–us-gov-west-11.206.002.400.1201.50200,000aggregator
DeepSeek V4 ProDeepInfrafp8+priority–1.953.902.440.150–1,048,576api
Mistral Medium 3.5OpenRouter––1.507.503.00––262,144api
Mistral Medium 3.5Mistral AIzdr–1.507.503.00––262,144api
Mistral Medium 3.5 3 sourcesMistral AI––1.507.503.000.150–262,144api

Confidence: official = vendor pricing page · api = OpenRouter per-host · aggregator = community list, may lag · unreviewed = automatic extraction awaiting review. When sources disagree the most authoritative wins. Details: /methodology#confidence