AI Model Prices Compared: Live Per-Token Costs for Coding
Prices update automatically as providers move them. Last synced 2026-09-28. Client prices include the platform margin; free-tier models cost 0.
| Model | Provider | Input $/1M | Output $/1M | Context |
|---|---|---|---|---|
| Mimo V2 Flash | mimo | $0.12 | $0.36 | 262K |
| GPT 6 Luna | openai | $0.15 | $0.75 | 1050K |
| Llama 4 Scout | groq | $0.165 | $0.51 | 1000K |
| DeepSeek V4 Flash | deepseek | $0.21 | $0.42 | 1000K |
| Mimo V2.5 | mimo | $0.21 | $0.42 | 131K |
| Mimo V2.6 Flash | mimo | $0.21 | $0.42 | 1000K |
| Qwen3.8 Flash | dashscope | $0.225 | $0.705 | 1000K |
| Mistral Small 4 | mistral | $0.225 | $0.9 | 128K |
| Hy3 | tencent | $0.225 | $0.885 | 32K |
| GLM 5.3 Flash | zai | $0.225 | $0.75 | 1000K |
| Qwen3.6 Flash | dashscope | $0.2812 | $1.6875 | 1000K |
| Qwen3.5 Plus | dashscope | $0.3 | $1.8 | 65K |
| GPT 5.6 Luna | openai | $0.3 | $1.8 | 1050K |
| GPT 5.4 Nano | openai | $0.3 | $1.875 | 400K |
| Gemini 3.1 Flash-Lite | $0.375 | $2.25 | 65K | |
| DeepSeek V4.1 Flash | deepseek | $0.45 | $1.8 | 1000K |
| MiniMax M2.5 | minimax | $0.45 | $1.8 | 1000K |
| MiniMax M2.7 | minimax | $0.45 | $1.8 | 204K |
| MiniMax M3 | minimax | $0.45 | $1.8 | 1000K |
| Mimo V2 Omni | mimo | $0.48 | $2.4 | 262K |
| Qwen3.7 Plus | dashscope | $0.6 | $2.4 | 1000K |
| Mimo V2.5 Pro | mimo | $0.6525 | $1.305 | 131K |
| Mimo V2.6 Pro | mimo | $0.6525 | $1.305 | 1000K |
| Qwen3.6 Plus | dashscope | $0.75 | $4.5 | 1000K |
| Gemini 3 Flash (Preview) | $0.75 | $4.5 | 65K | |
| Mistral Large 4 | mistral | $0.75 | $2.25 | 128K |
| DeepSeek R1 | deepseek | $0.825 | $3.285 | 128K |
| Kimi K2.5 | moonshot | $0.9 | $4.5 | 262K |
| GLM 4.7 | zai | $0.9 | $3.3 | 200K |
| GPT 5.4 Mini | openai | $1.125 | $6.75 | 128K |
| Mimo V2 Pro | mimo | $1.2 | $3.6 | 1000K |
| Kimi K2.6 | moonshot | $1.425 | $6 | 262K |
| Kimi K2.7 Code | moonshot | $1.425 | $6 | 262K |
| Claude Haiku 4.5 | anthropic | $1.5 | $7.5 | 64K |
| GLM 5 | zai | $1.5 | $4.8 | 200K |
| Qwen3 Max | dashscope | $1.8 | $9 | 65K |
| GLM 5.1 | zai | $2.1 | $6.6 | 202K |
| GLM 5.2 | zai | $2.1 | $6.6 | 1000K |
| GLM 5.3 | zai | $2.1 | $6.6 | 1000K |
| Gemini 3.5 Flash | $2.25 | $13.5 | 1000K | |
| Gemini 3.8 Flash | $2.25 | $11.25 | 1048K | |
| Gemini 3.7 Flash | $2.25 | $11.25 | 1048K | |
| DeepSeek V4 Pro | deepseek | $2.61 | $5.22 | 1000K |
| GPT 5.3 Codex | openai | $2.625 | $21 | 128K |
| Claude Sonnet 5 | anthropic | $3 | $15 | 64K |
| Qwen3.8 Max | dashscope | $3 | $9 | 1000K |
| Gemini 3.1 Pro | $3 | $18 | 65K | |
| GPT 6 Sol | openai | $3 | $15 | 1050K |
| GPT 5.6 Terra | openai | $3 | $18 | 1050K |
| Grok 4.5 | xai | $3 | $9 | 1000K |
| Grok 4.6 | xai | $3 | $9 | 500K |
| Grok 4.7 | xai | $3 | $9 | 500K |
| Qwen3.7 Max | dashscope | $3.75 | $11.25 | 1000K |
| GPT 5.4 | openai | $3.75 | $22.5 | 128K |
| Claude Sonnet 4.5 | anthropic | $4.5 | $22.5 | 64K |
| Claude Sonnet 4.6 | anthropic | $4.5 | $22.5 | 1000K |
| Kimi K3 | moonshot | $4.5 | $22.5 | 1048K |
| Claude Opus 5.5 | anthropic | $6 | $30 | 1000K |
| GPT 5.6 Sol | openai | $6 | $30 | 1050K |
| Claude Opus 4.8 | anthropic | $7.5 | $37.5 | 128K |
| Claude Opus 4.5 | anthropic | $7.5 | $37.5 | 200K |
| GPT 5.5 | openai | $7.5 | $45 | 128K |
| Claude Fable 5 | anthropic | $15 | $75 | 128K |
| Claude Fable 5.1 | anthropic | $15 | $75 | 1000K |
| GPT 6 Astra | openai | $15 | $75 | 1050K |
| GPT 5.4 Pro | openai | $45 | $270 | 128K |
| GPT 5.5 Pro | openai | $45 | $270 | 1050K |
| Kintio Auto | free | Free | Free | 128K |
How to read this table
Input price applies to everything you send, output price to everything generated. Agent loops multiply both, so blended cost per completed task matters more than either column. Cache-friendly workloads should weigh cache discounts, published separately by most providers.
Frequently asked questions
Why do prices change?
Providers reprice regularly; this page re-syncs with upstream pricing feeds.
Which model is cheapest for coding?
Sort by input price, then test on your tickets — the cheapest sticker price that completes work without supervision wins.
Are free models really free?
Yes. Free-tier rows cost 0 and are served through managed free routes with fallbacks.
Use these prices directly
One Kintio key serves every model on this page through Claude Code and other tools.
Explore Kintio →