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 ↑?ContextLast changeConfidence?
Llama 3.3 70B Instruct 2 sourcesDeepInfra––0.2300.4000.273128,0002026-08-27 ↓aggregator
Grok 4 FastAzure AI Foundry––0.2000.5000.2752,000,0002026-01-14 ↓aggregator
Llama 3.3 70B InstructAkashMLfp8–0.2000.5200.280128,000–api
Llama 4 Scout 3 sourcesNovita AI––0.1800.5900.28310,000,0002026-01-13 ↓api
Llama 4 ScoutTogether AI––0.1800.5900.28310,000,000–aggregator
Mistral Small 4Mistral AI–eu0.1650.6600.289262,144–api
Mistral Small 4Mistral AI–us0.1650.6600.289262,144–api
Llama 3.3 70B InstructParasailfp8–0.2200.5000.290128,000–api
DeepSeek V4 FlashAzure AI Foundry–us0.2100.5600.2981,048,576–api
Qwen3-235B-A22B 2 sourcesNebius AI Studio––0.2000.6000.300131,072–aggregator
Codestral 25.01Google Vertex AI––0.2000.6000.300262,144–aggregator
Llama 4 MaverickDigitalOcean––0.1880.6530.3041,000,000–api
Gemini 2.5 Flash-LiteGooglepriority–0.1800.7200.3151,048,576–api
DeepSeek V3.2 (chat)DeepSeek––0.2800.4200.315128,0002026-01-13 ↓aggregator
DeepSeek V3.2 (reasoner)DeepSeek––0.2800.4200.315128,0002026-01-13 ↓aggregator
GPT-5 miniOpenAIflex–0.1251.000.344400,000–api
Llama 4 Scout 2 sourcesAzure AI Foundry––0.2000.7800.34510,000,000–aggregator
Llama 4 Maverick 2 sourcesDeepInfra––0.2000.8000.3501,000,0002026-08-27 ↑aggregator
Qwen3-235B-A22B 2 sourcesNovita AI––0.2000.8000.350131,0722026-01-13 ↓aggregator
Llama 4 ScoutGoogle Vertex AI–us-east50.2500.7000.36310,000,000–api
Llama 4 Scout 2 sourcesGoogle Vertex AI––0.2500.7000.36310,000,000–aggregator
DeepSeek V3.2 (chat)DeepInfraflex+fp4–0.2560.7120.370128,000–api
Qwen3-235B-A22BFireworks AI––0.2200.8800.385131,072–aggregator
GPT-6 Luna 2 sourcesOpenAIfast–0.2001.000.4001,050,000–official
DeepSeek V3.2 (reasoner)Hyperbolic––0.4000.4000.400128,000–aggregator
Llama 4 Maverick 2 sourcesNovita AI––0.2700.8500.4151,000,0002026-01-13 ↓aggregator
Gemini 2.5 FlashGoogleflex–0.1501.250.4251,048,576–api
DeepSeek V3.2 (chat) 2 sourcesDeepInfra––0.2500.9500.425128,0002026-08-27 ↓aggregator
Llama 4 Maverick 2 sourcesAzure AI Foundry––0.2501.000.4381,000,0002026-09-17 ↕aggregator
Codestral 25.01 2 sourcesAzure AI Foundry––0.3000.9000.450262,144–aggregator
Codestral 25.01Mistral AI––0.3000.9000.450262,1442026-08-21 ↓aggregator
DeepSeek V3.2 (chat)OpenRouter––0.2571.030.450128,000–api
DeepSeek V3.2 (chat)DeepInfrafp4–0.3200.8900.463128,000–api
DeepSeek V4 FlashWafermxfp4–0.6000.1700.4931,048,5762026-10-27 ↓api
DeepSeek V3.2 (chat)StreamLake promo −10%––0.2861.140.500128,000–api
Grok 3 miniAzure AI Foundry––0.2501.270.505131,0722026-01-14 ↓aggregator
Llama 4 MaverickOpenRouter––0.3501.000.5131,000,000–api
Llama 4 MaverickParasailfp8–0.3501.000.5131,000,000–api
Llama 4 MaverickGoogle Vertex AI–us-east50.3501.150.5501,000,000–api
Llama 4 Maverick 2 sourcesGoogle Vertex AI––0.3501.150.5501,000,000–aggregator
Qwen3-Coder-480B-A35BGoogle Vertex AI––0.2201.800.615262,1442026-08-21 ↓aggregator
Inkling SmallOpenRouter––0.4501.200.638524,288–api
Inkling SmallDeepInfrafp8–0.4501.200.638524,288–api
Inkling Small 2 sourcesDeepInfra––0.4501.200.638524,288–aggregator
Inkling SmallDeepInfrafp4–0.4501.200.638524,288–api
Llama 3.3 70B Instruct 2 sourcesGroq––0.5900.7900.640128,000–api
Qwen3-Coder-480B-A35B 2 sourcesNovita AI––0.3801.550.673262,1442026-08-27 ↑aggregator
GPT-5 miniOpenRouter––0.2502.000.688400,000–api
GPT-5 mini 3 sourcesAzure AI Foundry––0.2502.000.688400,000–api
GPT-5 mini 3 sourcesOpenAI––0.2502.000.688400,000–api

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