Skip to main content
TokenCost logoTokenCost

Cheapest LLM Models in 2026

All 136+ LLM API models ranked from cheapest to most expensive by average token price. Find the most budget-friendly AI model for your project.

RankModelProviderInput/1MOutput/1MAvg/1MContext
#1Gemma 4 12BGoogle$0$0$0.00262K
#2DiffusionGemma 26BGoogle$0$0$0.00262K
#3Qwen3.5-9BAlibaba$0.05$0.15$0.10262K
#4GPT-OSS 120BOpenAI$0.039$0.19$0.11131K
#5Ministral 3 8BMistral$0.15$0.15$0.15262K
#6Pixtral 12BMistral$0.15$0.15$0.15128K
#7Laguna S 2.1Poolside$0.1$0.2$0.151.0M
#8Hunyuan HY3 PreviewTencent$0.066$0.26$0.16262K
#9Gemma 4 26B A4BGoogle$0.06$0.3$0.18262K
#10Llama 3.3 70BMeta$0.18$0.18$0.18131K
#11GPT-OSS 20B (Bedrock)OpenAI$0.07$0.3$0.1816K
#12GPT-OSS 20BOpenAI$0.075$0.3$0.19131K
#13Llama 4 ScoutMeta$0.08$0.3$0.191.0M
#14Devstral SmallMistral$0.1$0.3$0.20256K
#15Mistral Small 3.2Mistral$0.1$0.3$0.20128K
#16Ministral 3 14BMistral$0.2$0.2$0.20262K
#17DeepSeek V4-FlashDeepSeek$0.14$0.28$0.211.0M
#18GPT-5 NanoOpenAI$0.05$0.4$0.23128K
#19GLM-4.7-flashZhipu$0.07$0.4$0.24200K
#20Gemma 4 31BGoogle$0.12$0.36$0.24262K
#21GPT-4.1 NanoOpenAI$0.1$0.4$0.251.0M
#22Gemini 2.5 Flash-LiteGoogle$0.1$0.4$0.251.0M
#23DeepSeek V3.2 (Chat)DeepSeek$0.28$0.42$0.35128K
#24DeepSeek V3.2 (Reasoner)DeepSeek$0.28$0.42$0.35128K
#25GPT-OSS 120B (Bedrock)OpenAI$0.15$0.6$0.3816K
#26GPT-4o MiniOpenAI$0.15$0.6$0.38128K
#27Mistral Small 4Mistral$0.15$0.6$0.38256K
#28KAT-Coder-Air v2.5Kwaipilot$0.15$0.6$0.38262K
#29Qwen3-Next-80B-A3B-ThinkingAlibaba$0.0975$0.78$0.44262K
#30Qwen3 Coder NextAlibaba$0.11$0.8$0.46262K
#31Mercury 2Inception Labs$0.25$0.75$0.50128K
#32Nemotron 3 Super 120BNVIDIA$0.3$0.8$0.551.0M
#33Llama 4 MaverickMeta$0.27$0.85$0.561.0M
#34CodestralMistral$0.3$0.9$0.60256K
#35DeepSeek V4-ProDeepSeek$0.435$0.87$0.651.0M
#36MiMo-V2.5-ProXiaomi$0.435$0.87$0.651.0M
#37Step 3.7 FlashStepFun$0.2$1.15$0.67262K
#38GPT-5.4 NanoOpenAI$0.2$1.25$0.72400K
#39MiniMax M2.7MiniMax$0.3$1.2$0.75205K
#40MiniMax M2.5MiniMax$0.3$1.2$0.75128K
#41Gemini 3.1 Flash-LiteGoogle$0.25$1.5$0.881.0M
#42Qwen3.6-PlusAlibaba$0.276$1.651$0.961.0M
#43GPT-4.1 MiniOpenAI$0.4$1.6$1.001.0M
#44Mistral Large 3Mistral$0.5$1.5$1.00262K
#45Magistral SmallMistral$0.5$1.5$1.0040K
#46Qwen3.7 PlusAlibaba$0.4$1.6$1.001.0M
#47GPT-5 MiniOpenAI$0.25$2$1.13400K
#48DevstralMistral$0.4$2$1.20256K
#49Qwen3.5-Omni FlashAlibaba$0.4$2.2$1.30262K
#50Doubao Seed 2.1 TurboByteDance$0.44$2.21$1.32262K
#51Qwen 3.5 27BAlibaba$0.3$2.4$1.35128K
#52DeepSeek R1DeepSeek$0.55$2.19$1.37128K
#53Gemini 3.5 Flash-LiteGoogle$0.3$2.5$1.401.0M
#54Gemini 2.5 FlashGoogle$0.3$2.5$1.401.0M
#55Nova 2.0 LiteAmazon$0.3$2.5$1.401.0M
#56GLM-4.7Zhipu$0.6$2.2$1.40200K
#57Grok Build 0.1xAI$1$2$1.50256K
#58Nemotron 3 Ultra 550BNVIDIA$0.5$2.5$1.501.0M
#59MiniMax M3MiniMax$0.6$2.4$1.501.0M
#60Nex-N2-ProNex AGI$0.5$2.5$1.50262K
#61Kimi K2 ThinkingMoonshot$0.6$2.5$1.55262K
#62QwQ-PlusAlibaba$0.8$2.4$1.60131K
#63Gemini 3 FlashGoogle$0.5$3$1.751.0M
#64Gemini 3 Flash ReasoningGoogle$0.5$3$1.751.0M
#65Kimi K2.5Moonshot$0.6$3$1.80262K
#66LongCat-2.0Meituan$0.75$2.95$1.851.0M
#67KAT-Coder-Pro v2.5Kwaipilot$0.74$2.96$1.85262K
#68Grok 4.3xAI$1.25$2.5$1.881.0M
#69Grok 4.20xAI$1.25$2.5$1.881.0M
#70MiMo-V2-ProXiaomi$1$3$2.001.0M
#71Qwen 3.5 397BAlibaba$0.6$3.6$2.10128K
#72GLM-5Zhipu$1$3.2$2.10128K
#73Claude Haiku 3.5Anthropic$0.8$4$2.40200K
#74Kimi K2.7 CodeMoonshot$0.95$4$2.48262K
#75Kimi K2.6Moonshot$0.95$4$2.48262K
#76Qwen3.5-Omni PlusAlibaba$0.4$4.8$2.60262K
#77GLM-5 TurboZhipu$1.2$4$2.60200K
#78GPT-5.4 MiniOpenAI$0.75$4.5$2.63400K
#79Gemini 3.1 Flash LiveGoogle$0.75$4.5$2.631.0M
#80Doubao Seed 2.1 ProByteDance$0.88$4.41$2.65262K
#81o4 MiniOpenAI$1.1$4.4$2.75200K
#82o3 MiniOpenAI$1.1$4.4$2.75200K
#83Muse Spark 1.1Meta$1.25$4.25$2.751.0M
#84GLM-5.2Zhipu$1.4$4.4$2.901.0M
#85GLM-5.1Zhipu$1.4$4.4$2.90200K
#86Claude Haiku 4.5Anthropic$1$5$3.00200K
#87GPT-5.6 LunaOpenAI$1$6$3.501.1M
#88Magistral MediumMistral$2$5$3.5040K
#89Grok 4.5xAI$2$6$4.00500K
#90Pixtral LargeMistral$2$6$4.00128K
#91Gemini 3.6 FlashGoogle$1.5$7.5$4.501.0M
#92Mistral Medium 3.5Mistral$1.5$7.5$4.50256K
#93Qwen3.6-Max-PreviewAlibaba$1.3$7.8$4.55262K
#94Kimi K2 Thinking TurboMoonshot$1.15$8$4.58262K
#95GPT-4.1OpenAI$2$8$5.001.0M
#96o3OpenAI$2$8$5.00200K
#97o4 Mini Deep ResearchOpenAI$2$8$5.00200K
#98Qwen3.7 MaxAlibaba$2.5$7.5$5.001.0M
#99Gemini 3.5 FlashGoogle$1.5$9$5.251.0M
#100GPT-5.1OpenAI$1.25$10$5.63400K
#101GPT-5OpenAI$1.25$10$5.63400K
#102Gemini 2.5 ProGoogle$1.25$10$5.631.0M
#103Nova 2.0 Pro ReasoningAmazon$1.25$10$5.63128K
#104Claude Sonnet 5Anthropic$2$10$6.001.0M
#105GPT-4oOpenAI$2.5$10$6.25128K
#106Command A+Cohere$2.5$10$6.25128K
#107Command ACohere$2.5$10$6.25128K
#108Gemini 3.1 ProGoogle$2$12$7.001.0M
#109Gemini 3 ProGoogle$2$12$7.001.0M
#110GPT-5.2OpenAI$1.75$14$7.88400K
#111GPT-5.3 CodexOpenAI$1.75$14$7.88400K
#112Voxtral TTSMistral$16$0$8.00128K
#113GPT-5.6 TerraOpenAI$2.5$15$8.751.1M
#114GPT-5.4OpenAI$2.5$15$8.751.1M
#115Claude Sonnet 4.6Anthropic$3$15$9.001.0M
#116Claude Sonnet 4.5Anthropic$3$15$9.00200K
#117Claude 3.7 SonnetAnthropic$3$15$9.00200K
#118Sonar ProPerplexity$3$15$9.00128K
#119Kimi K3Moonshot$3$15$9.001.0M
#120Gemini 3.1 Flash TTSGoogle$1$20$10.5032K
#121Claude Opus 5Anthropic$5$25$15.001.0M
#122Claude Opus 4.8Anthropic$5$25$15.001.0M
#123Claude Opus 4.7Anthropic$5$25$15.001.0M
#124Claude Opus 4.6Anthropic$5$25$15.001.0M
#125Claude Opus 4.5Anthropic$5$25$15.00200K
#126GPT-5.6 SolOpenAI$5$30$17.501.1M
#127GPT-5.5OpenAI$5$30$17.501.1M
#128MAI-Image-2Microsoft$5$33$19.0032K
#129o3 Deep ResearchOpenAI$10$40$25.00200K
#130Claude Fable 5Anthropic$10$50$30.001.0M
#131o1OpenAI$15$60$37.50200K
#132Claude Opus 4.1Anthropic$15$75$45.00200K
#133o3-proOpenAI$20$80$50.00200K
#134GPT-5.5 ProOpenAI$30$180$105.001.1M
#135GPT-5.4 ProOpenAI$30$180$105.001.1M
#136o1 ProOpenAI$150$600$375.00200K

How We Rank the Cheapest LLMs

Models are ranked by their average price per 1 million tokens, calculated as (input price + output price) / 2. This gives a balanced view of overall cost since most workloads use both input and output tokens.

Keep in mind that the cheapest model isn't always the best choice. Consider quality benchmarks, context window size, and output speed when making your decision. Use our leaderboard to compare quality alongside cost.