Cheapest AI API per task and per 1M tokens
The lowest cost of the L2 task in our data is $0.0025 per task, on Ling 3.0 Tiny (prices as listed in the United States on ). The same three standard tasks at each model’s listed price, sorted by L2; model quality is not compared.
Cheapest per task 25 lowest by the L2 task
L1, L2 and L3 are three standard tasks. Each figure is the cost of one whole task at the model’s listed price, not a price per million tokens; L1 is shown per 1,000 tasks. L1: micro-task (classification or strict JSON extraction), 1,500 input and 200 output tokens. L2: deep analysis over a 50,000-token document, 3 steps. L3: agentic coding loop, 15 steps.
| Model | L1, per 1,000 tasks | L2, per task ↑ | L3, per task | Price from |
|---|---|---|---|---|
| Ling 3.0 TinyinclusionAI | $0.032 | $0.0025 | $0.020 | OpenRouter |
| Llama 3.1 8B InstructMeta | $0.038 | $0.0033 | $0.027 | OpenRouter |
| Ling 3.0 FlashinclusionAI | $0.044 | $0.0035 | $0.028 | OpenRouter |
| Nex-N2.5-MiniNex AGI | $0.058 | $0.0043 | $0.034 | OpenRouter |
| Mistral NemoMistral | $0.049 | $0.0047 | $0.038 | OpenRouter |
| Llama 3.2 1B InstructMeta | $0.081 | $0.0049 | $0.037 | OpenRouter |
| Qwen3.7 FlashQwen | $0.071 | $0.0052 | $0.041 | OpenRouter |
| Schematron V2 TurboInference.net | $0.075 | $0.0052 | $0.041 | OpenRouter |
| GLM 5.3 FlashZ.ai | $0.067 | $0.0056 | $0.045 | OpenRouter |
| Nova Micro 1.0Amazon | $0.081 | $0.0060 | $0.047 | OpenRouter |
| ExtractScaleDown | $0.060 | $0.0064 | $0.053 | OpenRouter |
| SummarizeScaleDown | $0.060 | $0.0064 | $0.053 | OpenRouter |
| Llama 3 8B LunarisSao10K | $0.070 | $0.0065 | $0.053 | OpenRouter |
| Mercury 2.5Inception | $0.090 | $0.0068 | $0.054 | OpenRouter |
| Hy-MT2-1.8BTencent | $0.101 | $0.0075 | $0.059 | OpenRouter |
| Ling 3.1 FlashinclusionAI | $0.095 | $0.0076 | $0.060 | OpenRouter |
| DeepSeek V4 Flash 0731DeepSeek | $0.125 | $0.0078 | $0.060 | OpenRouter |
| Mistral Small 3Mistral | $0.091 | $0.0082 | $0.066 | OpenRouter |
| Gemma 3 4BGoogle | $0.095 | $0.0083 | $0.067 | OpenRouter |
| Gemma 3 12BGoogle | $0.105 | $0.0084 | $0.067 | OpenRouter |
| Solar Mini 4Upstage | $0.115 | $0.0086 | $0.068 | OpenRouter |
| Schematron V2 SmallInference.net | $0.121 | $0.0086 | $0.068 | OpenRouter |
| Llama 3.2 3B InstructMeta | $0.141 | $0.0089 | $0.069 | OpenRouter |
| GPT-5 NanoOpenAI | $0.155 | $0.0092 | $0.069 | OpenRouter |
| Command R7B (12-2024)CohereNot independently verifiable | $0.086 | $0.0064 | $0.051 | OpenRouter |
Calculated L1: micro-task (classification or strict JSON extraction), 1,500 input and 200 output tokens; L2: deep analysis over a 50,000-token document in 3 steps; L3: agentic coding loop, 15 steps. Visible output only, no cache discount, no hidden reasoning; tokenizer and retries 1.00. Assumptions and formula
Cheapest from the larger AI labs OpenAI, Anthropic, Google, DeepSeek, Qwen, xAI, Mistral, Meta, Moonshot, Z.ai
| Model | L1, per 1,000 tasks | L2, per task ↑ | L3, per task | Price from |
|---|---|---|---|---|
| Llama 3.1 8B InstructMeta | $0.038 | $0.0033 | $0.027 | OpenRouter |
| Mistral NemoMistral | $0.049 | $0.0047 | $0.038 | OpenRouter |
| Llama 3.2 1B InstructMeta | $0.081 | $0.0049 | $0.037 | OpenRouter |
| Qwen3.7 FlashQwen | $0.071 | $0.0052 | $0.041 | OpenRouter |
| GLM 5.3 FlashZ.ai | $0.067 | $0.0056 | $0.045 | OpenRouter |
| DeepSeek V4 Flash 0731DeepSeek | $0.125 | $0.0078 | $0.060 | OpenRouter |
| Mistral Small 3Mistral | $0.091 | $0.0082 | $0.066 | OpenRouter |
| Gemma 3 4BGoogle | $0.095 | $0.0083 | $0.067 | OpenRouter |
| Gemma 3 12BGoogle | $0.105 | $0.0084 | $0.067 | OpenRouter |
| Llama 3.2 3B InstructMeta | $0.141 | $0.0089 | $0.069 | OpenRouter |
| GPT-5 NanoOpenAI | $0.155 | $0.0092 | $0.069 | OpenRouter |
| DeepSeek V4.1 Flash 0910DeepSeek | $0.190 | $0.0098 | $0.072 | OpenRouter |
| GLM 4.7 FlashZ.ai | $0.171 | $0.011 | $0.083 | OpenRouter |
| Qwen3.5-FlashQwen | $0.149 | $0.011 | $0.088 | OpenRouter |
| Gemma 4 26B A4BGoogle | $0.146 | $0.011 | $0.091 | OpenRouter |
| Mistral Small 3.2 24BMistral | $0.152 | $0.013 | $0.100 | OpenRouter |
| gpt-oss-20bOpenAI | $0.172 | $0.013 | $0.101 | Groq price list |
| gpt-oss-safeguard-20bOpenAI | $0.172 | $0.013 | $0.101 | OpenRouter |
| Qwen3 32BQwen | $0.176 | $0.014 | $0.108 | OpenRouter |
| Gemma 3 27BGoogle | $0.210 | $0.014 | $0.109 | OpenRouter |
| DeepSeek V4 Flash 0423DeepSeek | $0.173 | $0.015 | $0.121 | OpenRouter |
| Ministral 3 3B 2512Mistral | $0.170 | $0.016 | $0.132 | OpenRouter |
| Muse Spark 1.2 ContributorMeta | $0.190 | $0.017 | $0.133 | OpenRouter |
| Muse Spark 1.3 ContributorMeta | $0.190 | $0.017 | $0.133 | OpenRouter |
| Qwen2.5 7B InstructQwen | $0.190 | $0.017 | $0.133 | OpenRouter |
Lowest input price per 1M tokens
| Model | Input | Output | Price from |
|---|---|---|---|
| GLM 5.2Z.ai | $0.006 | $6.80 | OpenRouter |
| Ling 3.0 TinyinclusionAI | $0.015 | $0.045 | OpenRouter |
| Llama 3.1 8B InstructMeta | $0.02 | $0.04 | OpenRouter |
| Ling 3.0 FlashinclusionAI | $0.021 | $0.063 | OpenRouter |
| Nex-N2.5-MiniNex AGI | $0.025 | $0.10 | OpenRouter |
| Llama 3.2 1B InstructMeta | $0.027 | $0.201 | OpenRouter |
| Mistral NemoMistral | $0.029 | $0.03 | OpenRouter |
| Qwen3.7 FlashQwen | $0.03 | $0.13 | OpenRouter |
| Schematron V2 TurboInference.net | $0.03 | $0.15 | OpenRouter |
| GLM 5.3 FlashZ.ai | $0.0335 | $0.083 | OpenRouter |
| Nova Micro 1.0Amazon | $0.035 | $0.14 | OpenRouter |
| ExtractScaleDown | $0.04 | $0.00 | OpenRouter |
| SummarizeScaleDown | $0.04 | $0.00 | OpenRouter |
| Llama 3 8B LunarisSao10K | $0.04 | $0.05 | OpenRouter |
| Command R7B (12-2024)CohereNot independently verifiable | $0.0375 | $0.15 | OpenRouter |
Published USD per 1M tokens, prices as listed in the United States on .
Lowest output price per 1M tokens
| Model | Output | Input | Price from |
|---|---|---|---|
| Mistral NemoMistral | $0.03 | $0.029 | OpenRouter |
| Llama 3.1 8B InstructMeta | $0.04 | $0.02 | OpenRouter |
| Ling 3.0 TinyinclusionAI | $0.045 | $0.015 | OpenRouter |
| Llama 3 8B LunarisSao10K | $0.05 | $0.04 | OpenRouter |
| Ling 3.0 FlashinclusionAI | $0.063 | $0.021 | OpenRouter |
| Mistral Small 3Mistral | $0.08 | $0.05 | OpenRouter |
| GLM 5.3 FlashZ.ai | $0.083 | $0.0335 | OpenRouter |
| Nex-N2.5-MiniNex AGI | $0.10 | $0.025 | OpenRouter |
| Gemma 3 4BGoogle | $0.10 | $0.05 | OpenRouter |
| Ministral 3 3B 2512Mistral | $0.10 | $0.10 | OpenRouter |
| Reka Edgerekaai | $0.10 | $0.10 | OpenRouter |
| MythoMax 13Bgryphe | $0.11 | $0.08 | OpenRouter |
| Qwen3.7 FlashQwen | $0.13 | $0.03 | OpenRouter |
| Ternary Bonsai 2 9BPrismML | $0.13 | $0.08 | OpenRouter |
| Ling 3.1 FlashinclusionAI | $0.135 | $0.045 | OpenRouter |
Published USD per 1M tokens, prices as listed in the United States on .
Left out of these rankings: batch variants, free listings, models no longer listed, models with fewer than two price checks on two dates, and models whose name marks image, audio or music output. In each table, rows of providers whose terms restrict independent comparisons are shown last and greyed (list of these providers).
All API prices · Long-context cliff · Caching and batch · Rate limits by tier · What breaks when you switch · Rate limit planner · Task cost calculator · All models on OpenRouter · How we check prices