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 every active offering, including the long tail. Show curated models only

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

ModelHostVariantRegionInputOutputBlended ↑?Cache readCache writeBatch inBatch outContextConfidence?
GPT-5 nanoOpenRouter––0.05000.4000.1380.005000–––400,000api
Qwen3.5-9BParasail––0.1000.2500.138––––262,144api
GPT-5 nano 3 sourcesAzure AI Foundry––0.05000.4000.1380.0100–0.02500.200400,000api
GPT-5 nano 3 sourcesOpenAI––0.05000.4000.1380.005000–0.02500.200400,000api
Gemma 4 26B A4B 3 sourcesDeepInfra––0.07000.3400.138––––262,144official
Gemma 4 26B A4BMakora––0.09000.2800.1380.0490–––262,144api
GPT-5 nanoPerplexity––0.05000.4000.1380.005000–––400,000aggregator
Gemma 3 27B 3 sourcesNovita AI––0.1190.2000.139––––131,072api
Phi 4 MultimodalAzure AI Foundry––0.08000.3200.140––––131,072aggregator
Gemma 3 12B 2 sourcesAWS Bedrock––0.09000.2900.140––0.05000.150131,072aggregator
Hermes 2 Pro Llama 3 8B 2 sourcesNovita AI––0.1400.1400.140––––8,192aggregator
Phi-4-multimodalAzure AI Foundry––0.08000.3200.140––––128,000aggregator
Gemma 4 26B A4BNextBit promo −15%––0.09000.3000.1430.0500–––262,144api
Qwen3 30B A3B Instruct 2507DekaLLM––0.09000.3000.143––––262,144api
Qwen3 30B A3B Instruct 2507SiliconFlowfp8–0.09000.3000.143––––262,144api
Mistral Small 3.2 24BParasail––0.09000.3000.1430.0500–––256,000api
Llama Guard 4 12BDeepInfraflex–0.1440.1440.144––––163,840api
Gemma 3 12BAWS Bedrock–ap-southeast-20.09270.2990.144––––131,072aggregator
Hy3Tencent Cloudfp8+off-peak–0.08250.3300.1440.0206–––262,144api
Hy3OpenRouteroff-peak–0.08250.3300.1440.0206–––262,144api
MiMo-V2.6-Flashio.netfp8–0.1000.2800.1450.003000–––1,050,000api
MiMo-V2.6-FlashDarkbloomfp4–0.1000.2800.1450.0500–––1,050,000api
GLM 4.7 Flash 2 sourcesDeepInfra––0.06000.4000.1450.0100–––200,000aggregator
gpt-oss-120bSAIL Researchfp4–0.06000.4000.1450.0300–––131,072api
GLM 4.7 FlashOpenRouter––0.06050.4000.145––––200,000api
GLM 4.7 FlashCloudflare Workers AI––0.06050.4000.145––––200,000api
DeepSeek V4.1 Flash 1017ambientfp4–0.09800.2940.1470.002900–––1,048,576api
Gemma 3 27BNextBitfp8–0.08000.3500.1480.0400–––131,072api
gpt-oss-20bAWS Bedrock–ap-south-10.08000.3500.148––––131,072aggregator
MiMo-V2.5OpenRouter––0.1190.2380.1490.002550–––1,000,000api
DeepSeek V4 Flash 0731DigitalOcean––0.1190.2380.1490.0238–––1,048,576api
Ministral 3 8B 2512OpenRouter––0.1500.1500.1500.0150–––262,144api
Voxtral Small 24B 2507OpenRouter––0.1000.3000.1500.0100–––32,768api
Qwen3 30B A3B Instruct 2507OpenRouter––0.1000.3000.1500.100–––262,144api
Qwen3 14BOpenRouter––0.1200.2400.150––––40,960api
Llama 4 ScoutOpenRouter––0.1000.3000.150––––10,000,000api
DeepSeek V4 FlashRelacefp4–0.05000.4500.1500.0500–––1,048,576api
Step 3.5 FlashSiliconFlowfp8–0.1000.3000.150––––262,144api
Ministral 3 8B 2512Mistral AIzdr–0.1500.1500.1500.0150–––262,144api
Ministral 3 8B 2512 3 sourcesMistral AI––0.1500.1500.1500.0150–––262,144api
Voxtral Small 24B 2507Mistral AIzdr–0.1000.3000.1500.0100–––32,768api
Voxtral Small 24B 2507Mistral AI––0.1000.3000.1500.0100–––32,768api
gpt-oss-120bDigitalOcean––0.06000.4200.1500.0120–––131,072api
Qwen3 30B A3B Instruct 2507Nebius AI Studiofp8–0.1000.3000.1500.100–––262,144api
Mistral Small 3.2 24BMistral AI–eu0.1000.3000.1500.0100–––256,000api
Qwen3 14BDeepInfrafp8–0.1200.2400.150––––40,960api
Llama 4 ScoutDeepInfrafp8–0.1000.3000.150––––10,000,000api
Gemma 3 27BNebius AI Studiofp8–0.1000.3000.1500.0400–––131,072api
Mistral NemoMistral AI–eu0.1500.1500.1500.0150–––131,072api
Mistral NemoAzure AI Foundry––0.1500.1500.150––––131,072aggregator

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