LLM API Pricing Comparison (per 1M tokens)
Input and output prices for 43 models across 10 API hosts, with a calculator for your monthly bill. Sort by any column, or enter your own token volumes.
Prices updated · Source: OpenRouter models API · refreshed daily
Price table and cost calculator
| Host | Context | ||||
|---|---|---|---|---|---|
| Llama 3.1 8BMeta (open weights) | Groq | $0.050 | $0.080 | 131K | $0.660 |
| gpt-oss-20bOpenAI (open weights) | Groq | $0.075 | $0.30 | 131K | $1.35 |
| GPT-6 LunaOpenAI | OpenAI | $0.10 | $0.50 | 1.05M | $2.00 |
| Claude Haiku 5.5Anthropic | Anthropic | $0.10 | $0.50 | 1M | $2.00 |
| DeepSeek V4.1 FlashDeepSeek | DeepSeek | $0.15 | $0.60 | 1.05M | $2.70 |
| Mistral Small (2603)Mistral | Mistral | $0.15 | $0.60 | 262K | $2.70 |
| gpt-oss-120bOpenAI (open weights) | Groq | $0.15 | $0.60 | 131K | $2.70 |
| gpt-oss-120bOpenAI (open weights) | Together | $0.15 | $0.60 | 131K | $2.70 |
| GPT-5.4 nanoOpenAI | OpenAI | $0.20 | $1.25 | 400K | $4.50 |
| Codestral (2508)Mistral | Mistral | $0.30 | $0.90 | 256K | $4.80 |
| DeepSeek V4.1 FlashDeepSeek (open weights) | Together | $0.30 | $1.20 | 1.05M | $5.40 |
| Gemini 3.1 Flash-LiteGoogle | Google AI Studio | $0.25 | $1.50 | 1.05M | $5.50 |
| Command A PlusCohere | Cohere | $0.30 | $1.50 | 192K | $6.00 |
| Llama 3.3 70BMeta (open weights) | Groq | $0.59 | $0.79 | 131K | $7.48 |
| Gemini 3.5 Flash-LiteGoogle | Google AI Studio | $0.30 | $2.50 | 1.05M | $8.00 |
| Gemini 2.5 FlashGoogle | Google AI Studio | $0.30 | $2.50 | 1.05M | $8.00 |
| DeepSeek V4 ProDeepSeek | DeepSeek | $0.66 | $1.98 | 1.05M | $10.56 |
| Mistral Large 4Mistral | Mistral | $0.68 | $2.09 | 1.05M | $10.98 |
| Llama 3.3 70BMeta (open weights) | Together | $1.04 | $1.04 | 131K | $12.48 |
| Grok Build 0.1xAI | xAI | $1.00 | $2.00 | 256K | $14.00 |
| Gemini 3.8 FlashGoogle | Google AI Studio | $0.75 | $3.75 | 1.05M | $15.00 |
| GPT-5.4 miniOpenAI | OpenAI | $0.75 | $4.50 | 400K | $16.50 |
| Grok 4.3xAI | xAI | $1.25 | $2.50 | 1M | $17.50 |
| Grok 4.20xAI | xAI | $1.25 | $2.50 | 2M | $17.50 |
| Claude Haiku 4.5Anthropic | Anthropic | $1.00 | $5.00 | 200K | $20.00 |
| Mistral Medium 3.5Mistral | Mistral | $1.50 | $7.50 | 262K | $30.00 |
| Grok 4.7xAI | xAI | $2.00 | $6.00 | 500K | $32.00 |
| Qwen 3.8 MaxAlibaba | Alibaba | $2.00 | $6.00 | 1M | $32.00 |
| Gemini 2.5 ProGoogle | Google AI Studio | $1.25 | $10.00 | 1.05M | $32.50 |
| Gemini 3.5 FlashGoogle | Google AI Studio | $1.50 | $9.00 | 1.05M | $33.00 |
| GPT-6.1 SolOpenAI | OpenAI | $2.00 | $10.00 | 1.05M | $40.00 |
| Claude Sonnet 5.5Anthropic | Anthropic | $2.00 | $10.00 | 1M | $40.00 |
| Gemini 3.1 Pro (preview)Google | Google AI Studio | $2.00 | $12.00 | 1.05M | $44.00 |
| Command ACohere | Cohere | $2.50 | $10.00 | 256K | $45.00 |
| Kimi K3Moonshot (open weights) | Together | $2.70 | $13.50 | 1.05M | $54.00 |
| GPT-5.4OpenAI | OpenAI | $2.50 | $15.00 | 1.05M | $55.00 |
| Claude Sonnet 4.6Anthropic | Anthropic | $3.00 | $15.00 | 1M | $60.00 |
| Claude Opus 5.5Anthropic | Anthropic | $4.00 | $20.00 | 1M | $80.00 |
| Claude Opus 4.8Anthropic | Anthropic | $5.00 | $25.00 | 1M | $100.00 |
| GPT-5.5OpenAI | OpenAI | $5.00 | $30.00 | 1.05M | $110.00 |
| GPT-6 AstraOpenAI | OpenAI | $10.00 | $50.00 | 1.05M | $200.00 |
| Claude Fable 5.1Anthropic | Anthropic | $10.00 | $50.00 | 1M | $200.00 |
| GPT-5.5 ProOpenAI | OpenAI | $30.00 | $180.00 | 1.05M | $660.00 |
Sponsor slot
Sponsor this table
Reach developers comparing LLM API prices. One sponsor at a time, clearly labelled.
Head-to-head price comparisons
- GPT-6.1 Sol vs Claude Sonnet 5.5 pricing
- GPT-6.1 Sol vs Claude Opus 5.5 pricing
- GPT-6.1 Sol vs Gemini 3.1 Pro (preview) pricing
- Claude Sonnet 5.5 vs Gemini 3.1 Pro (preview) pricing
- Claude Opus 5.5 vs Gemini 3.1 Pro (preview) pricing
- GPT-6 Luna vs Claude Haiku 5.5 pricing
- Claude Haiku 5.5 vs Gemini 3.8 Flash pricing
- GPT-6.1 Sol vs Grok 4.7 pricing
- DeepSeek V4 Pro vs GPT-6.1 Sol pricing
- DeepSeek V4 Pro vs Claude Sonnet 5.5 pricing
Notes on reading the table
- Standard tier only. Batch APIs (typically about half price), flex and priority tiers, and prompt-cache discounts are not shown. Several providers also charge more for very long prompts.
- DeepSeek bills at double the listed rate during its peak hours (UTC), per its own pricing page. The table shows the off-peak rate.
- Open-weight models (gpt-oss, Llama, DeepSeek, Kimi) are sold by several hosts at different prices. Groq and Together rows show those hosts' own prices for the same weights.
- Prices change. This page re-reads the source daily; check the provider's pricing page before committing to a budget.
FAQ
What is the cheapest LLM API?
For a workload of 10M input and 2M output tokens a month, the cheapest model in this table is Llama 3.1 8B on Groq, at $0.050 per 1M input tokens and $0.080 per 1M output tokens. Among models sold by the lab that trained them, it is GPT-6 Luna (OpenAI). Rankings change with your input:output ratio, so use the calculator above with your own numbers.
Which LLM API has the most expensive output tokens?
GPT-5.5 Pro on OpenAI, at $180.00 per 1M output tokens ($30.00 per 1M input).
How is LLM API price per token calculated?
Providers bill input (prompt) tokens and output (completion) tokens separately, quoted per 1 million tokens. Monthly cost = input tokens (millions) × input price + output tokens (millions) × output price. Output tokens usually cost 3–8× more than input, so output-heavy workloads such as long generations or reasoning models are dominated by the output price.
Where do these prices come from?
From OpenRouter's public models API, refreshed daily. Each row is the standard-tier price that one named host publishes there; batch, flex and priority tiers are excluded. A model is left out rather than guessed when its host has no standard-tier listing. Last refresh: October 10, 2026.
Is the API up right now?
A cheap API you can't reach costs more than an expensive one that answers. Live status for these providers: