AI API pricing comparison
Price per 1M tokens for 29 models from OpenAI, Anthropic, Google, Meta, DeepSeek, Qwen, Mistral and xAI. Checked Oct 16, 2026
| Provider | Source | ||||||
|---|---|---|---|---|---|---|---|
| GPT-6 Astragpt-6-astraPrompts up to 272K. Longer prompts cost $20 in, $75 out. | OpenAI | $10 | $1 | $50 | $20 | 1.05M | Official |
| GPT-6.1 Solgpt-6.1-solPrompts up to 272K. Longer prompts cost $4 in, $15 out. | OpenAI | $2 | $0.10 | $10 | $4 | 1.05M | Official |
| GPT-6 Solgpt-6-solPrompts up to 272K. | OpenAI | $2 | $0.20 | $10 | $4 | 1.05M | Official |
| GPT-6 Lunagpt-6-lunaPrompts up to 272K. | OpenAI | $0.10 | $0.01 | $0.50 | $0.20 | 1.05M | Official |
| GPT-5.6 Solgpt-5.6-solPromotional price, at least through Nov 21, 2026. | OpenAI | $4 | $0.40 | $20 | $8 | n/a | Official |
| GPT-5.6 Terragpt-5.6-terraPrompts up to 272K. | OpenAI | $2 | $0.20 | $12 | $4.50 | n/a | Official |
| GPT-5.5gpt-5.5Prompts up to 272K. | OpenAI | $5 | $0.50 | $30 | $11.25 | n/a | Official |
| Claude Fable 5.1claude-fable-5.1Cache reads $0.25, 5-minute cache writes $12.50. | Anthropic | $10 | $0.25 | $50 | $20 | 1M | Official |
| Claude Opus 5.5claude-opus-5.55-minute cache writes $5. | Anthropic | $4 | $0.20 | $20 | $8 | 1M | Official |
| Claude Sonnet 5.5claude-sonnet-5.55-minute cache writes $2.50. | Anthropic | $2 | n/a | $10 | $4 | 1M | Official |
| Claude Haiku 5.5claude-haiku-5.5Rate for prompts up to 100K tokens. | Anthropic | $0.50 | n/a | $2.50 | $1 | 1M | Official |
| Claude Opus 5claude-opus-5Legacy model. | Anthropic | $5 | $0.50 | $25 | $10 | 1M | Official |
| Gemini 4 Argongemini-4-argonLaunch price through Nov 30, 2026. From Dec 1: $4 in, $20 out. | $2 | $0.10 | $10 | $4 | 1M | Official | |
| Gemini 3.8 Flashgemini-3.8-flashThrough Dec 31, 2026. From Jan 1, 2027: $1.50 in, $7.50 out. | $0.75 | $0.075 | $3.75 | $1.50 | 1M | Official | |
| Gemini 3.1 Pro Previewgemini-3.1-pro-previewPrompts up to 200K. Longer prompts cost $4 in, $18 out. | $2 | $0.20 | $12 | $4.50 | n/a | Official | |
| Gemini 3.5 Flash-Litegemini-3.5-flash-lite | $0.30 | $0.03 | $2.50 | $0.85 | n/a | Official | |
| Muse Wonder 1.0muse-wonder-1.0No long-context premium. | Meta | $4 | $0.20 | $20 | $8 | 1M | Official |
| Muse Spark 1.3muse-spark-1.3Contributor mode (trains on your data): $0.10 in, $0.20 out. | Meta | $1.25 | $0.15 | $4.25 | $2 | 1M | Official |
| DeepSeek V4.1 Flashdeepseek-flashPeak rate. Off-peak is half price. | DeepSeek | $0.30 | $0.006 | $1.20 | $0.525 | 1M | Official |
| DeepSeek V4 Prodeepseek-v4-proPeak rate. Off-peak is half price. | DeepSeek | $1.32 | $0.044 | $3.96 | $1.98 | 1M | Official |
| Qwen3.8 Maxqwen3.8-maxInternational (Singapore). Cache hits billed at 10% of input. | Qwen | $2 | $0.20 | $6 | $3 | 1M | Official |
| Qwen3.7 Plusqwen3.7-plusList price, prompts up to 256K. Limited-time 20% off. | Qwen | $0.40 | $0.04 | $1.60 | $0.70 | 1M | Official |
| Qwen3.8 Flashqwen3.8-flashInternational (Singapore). Cache hits billed at 10% of input. | Qwen | $0.15 | $0.015 | $0.47 | $0.23 | 1M | Official |
| Mistral Large 4mistral-large-4Sale price. Regular: $1.36 in, $4.18 out. | Mistral | $0.68 | $0.07 | $2.09 | $1.03 | 1M | Official |
| Mistral Medium 3.5mistral-medium-3.5 | Mistral | $1.50 | $0.15 | $7.50 | $3 | 256K | Official |
| Mistral Small 4mistral-small-4 | Mistral | $0.15 | $0.015 | $0.60 | $0.2625 | 256K | Official |
| Grok 4.7grok-4.7Prompts under 200K. Longer prompts cost $4 in, $12 out. | xAI | $2 | $0.50 | $6 | $3 | 500K | Official |
| Grok 4.3grok-4.3Prompts under 200K. Longer prompts cost double. | xAI | $1.25 | $0.20 | $2.50 | $1.56 | 1M | Official |
| Grok Build 0.1grok-build-0.1Prompts under 200K. | xAI | $1 | $0.20 | $2 | $1.25 | 256K | Official |
How to read these prices
Input is what you pay for the prompt you send, output is what the model writes back, and cached input is the discounted rate when part of your prompt repeats. Output usually costs 3 to 8 times more than input, so the 3:1 blend gives a fair single number for typical chat and coding use. Many providers charge more for very long prompts and less for batch jobs; those details are in each model's note.
Cheapest LLM APIs right now
By blended price the cheapest models are GPT-6 Luna ($0.20), Qwen3.8 Flash ($0.23), Mistral Small 4 ($0.2625). For top-end work, 13 of the 29 models charge $10 or more per 1M output tokens. If you use a chat app all day, a flat plan may still be cheaper: compare plans side by side or check our plan guides.
OpenRouter vs direct pricing
We list direct prices from each provider's own pricing page. OpenRouter resells most of these models through one API and usually passes through the provider's list price, but its own credit fees and per-provider routing can change what you pay. We don't copy OpenRouter's prices, since their terms don't allow automated copying, so check their model page before you commit.
Sources
- OpenAI pricing
- Anthropic pricing
- Google pricing
- Meta pricing
- DeepSeek pricing
- Qwen pricing
- Mistral pricing
- xAI pricing
FAQ
What is the cheapest AI API?
GPT-6 Luna from OpenAI is the cheapest in this table at $0.10 per 1M input tokens and $0.50 per 1M output tokens.
How much does the Claude API cost?
Claude Fable 5.1 $10 in, $50 out; Claude Opus 5.5 $4 in, $20 out; Claude Sonnet 5.5 $2 in, $10 out; Claude Haiku 5.5 $0.50 in, $2.50 out; Claude Opus 5 $5 in, $25 out per 1M tokens.
How much does the OpenAI API cost?
GPT-6 Astra $10 in, $50 out; GPT-6.1 Sol $2 in, $10 out; GPT-6 Sol $2 in, $10 out; GPT-6 Luna $0.10 in, $0.50 out; GPT-5.6 Sol $4 in, $20 out; GPT-5.6 Terra $2 in, $12 out; GPT-5.5 $5 in, $30 out per 1M tokens.
How much does the Gemini API cost?
Gemini 4 Argon $2 in, $10 out; Gemini 3.8 Flash $0.75 in, $3.75 out; Gemini 3.1 Pro Preview $2 in, $12 out; Gemini 3.5 Flash-Lite $0.30 in, $2.50 out per 1M tokens. Several Gemini prices are launch prices that rise later; see the notes.
How often are these prices checked?
Every day against each provider's official pricing page. The date at the top shows the last check.
Pricing by provider: OpenAI · Anthropic · Google · Meta · DeepSeek · Qwen · Mistral · xAI