Tools · Prompt caching
What cached input actually costs, model by model.
Cache read and cache write prices for 181 models, from OpenRouter's routed listings retrieved July 22, 2026. Sorted by the cached-input discount. Platform prices, not first-party provider prices; a bigger discount says nothing about quality.
| Model | Input $/1M | Cache read $/1M | Read cost vs input | Cache write $/1M |
|---|---|---|---|---|
| Xiaomi: MiMo-V2.5-Pro | $0.43 | $0.0036 | 1% | not listed |
| DeepSeek: DeepSeek V4 Pro | $0.43 | $0.0036 | 1% | not listed |
| Xiaomi: MiMo-V2.5 | $0.14 | $0.0028 | 2% | not listed |
| Meituan: LongCat 2.0 | $0.30 | $0.0060 | 2% | not listed |
| Anthropic Claude Haiku Latest | $1.00 | $0.10 | 10% | $1.25 |
| Anthropic Claude Sonnet Latest | $2.00 | $0.20 | 10% | $2.50 |
| Anthropic: Claude Fable 5 | $10.00 | $1.00 | 10% | $12.50 |
| Anthropic: Claude Fable Latest | $10.00 | $1.00 | 10% | $12.50 |
| Anthropic: Claude Haiku 4.5 | $1.00 | $0.10 | 10% | $1.25 |
| Anthropic: Claude Opus 4 | $15.00 | $1.50 | 10% | $18.75 |
| Anthropic: Claude Opus 4.1 | $15.00 | $1.50 | 10% | $18.75 |
| Anthropic: Claude Opus 4.5 | $5.00 | $0.50 | 10% | $6.25 |
| Anthropic: Claude Opus 4.6 | $5.00 | $0.50 | 10% | $6.25 |
| Anthropic: Claude Opus 4.7 | $5.00 | $0.50 | 10% | $6.25 |
| Anthropic: Claude Opus 4.7 (Fast) | $30.00 | $3.00 | 10% | $37.50 |
| Anthropic: Claude Opus 4.8 | $5.00 | $0.50 | 10% | $6.25 |
| Anthropic: Claude Opus 4.8 (Fast) | $10.00 | $1.00 | 10% | $12.50 |
| Anthropic: Claude Opus Latest | $5.00 | $0.50 | 10% | $6.25 |
| Anthropic: Claude Sonnet 4 | $3.00 | $0.30 | 10% | $3.75 |
| Anthropic: Claude Sonnet 4.5 | $3.00 | $0.30 | 10% | $3.75 |
| Anthropic: Claude Sonnet 4.6 | $3.00 | $0.30 | 10% | $3.75 |
| Anthropic: Claude Sonnet 5 | $2.00 | $0.20 | 10% | $2.50 |
| Google Gemini Flash Latest | $1.50 | $0.15 | 10% | $0.083 |
| Google Gemini Pro Latest | $2.00 | $0.20 | 10% | $0.38 |
| Google: Gemini 2.5 Flash | $0.30 | $0.030 | 10% | $0.083 |
| Google: Gemini 2.5 Flash Lite | $0.10 | $0.010 | 10% | $0.083 |
| Google: Gemini 2.5 Pro | $1.25 | $0.13 | 10% | $0.38 |
| Google: Gemini 2.5 Pro Preview 05-06 | $1.25 | $0.13 | 10% | $0.38 |
| Google: Gemini 2.5 Pro Preview 06-05 | $1.25 | $0.13 | 10% | $0.38 |
| Google: Gemini 3 Flash Preview | $0.50 | $0.050 | 10% | $0.083 |
| Google: Gemini 3.1 Flash Lite | $0.25 | $0.025 | 10% | $0.083 |
| Google: Gemini 3.1 Flash Lite Preview | $0.25 | $0.025 | 10% | $0.083 |
| Google: Gemini 3.1 Pro Preview | $2.00 | $0.20 | 10% | $0.38 |
| Google: Gemini 3.1 Pro Preview Custom Tools | $2.00 | $0.20 | 10% | $0.38 |
| Google: Gemini 3.5 Flash | $1.50 | $0.15 | 10% | $0.083 |
| Google: Gemini 3.5 Flash-Lite | $0.30 | $0.030 | 10% | $0.083 |
| Google: Gemini 3.6 Flash | $1.50 | $0.15 | 10% | $0.083 |
| Google: Nano Banana (Gemini 2.5 Flash Image) | $0.30 | $0.030 | 10% | $0.083 |
| Google: Nano Banana Pro (Gemini 3 Pro Image Preview) | $2.00 | $0.20 | 10% | $0.38 |
| Google: Nano Banana Pro (Gemini 3 Pro Image) | $2.00 | $0.20 | 10% | $0.38 |
| Inception: Mercury 2 | $0.25 | $0.025 | 10% | not listed |
| MiniMax: MiniMax M2 | $0.30 | $0.030 | 10% | not listed |
| MiniMax: MiniMax M2-her | $0.30 | $0.030 | 10% | not listed |
| MiniMax: MiniMax M2.1 | $0.30 | $0.030 | 10% | not listed |
| Mistral Large | $2.00 | $0.20 | 10% | not listed |
| Mistral Large 2407 | $2.00 | $0.20 | 10% | not listed |
| Mistral: Codestral 2508 | $0.30 | $0.030 | 10% | not listed |
| Mistral: Devstral 2 2512 | $0.40 | $0.040 | 10% | not listed |
| Mistral: Ministral 3 14B 2512 | $0.20 | $0.020 | 10% | not listed |
| Mistral: Ministral 3 3B 2512 | $0.10 | $0.010 | 10% | not listed |
| Mistral: Ministral 3 8B 2512 | $0.15 | $0.015 | 10% | not listed |
| Mistral: Mistral Large 3 2512 | $0.50 | $0.050 | 10% | not listed |
| Mistral: Mistral Medium 3 | $0.40 | $0.040 | 10% | not listed |
| Mistral: Mistral Medium 3.1 | $0.40 | $0.040 | 10% | not listed |
| Mistral: Mistral Small 3.2 24B | $0.10 | $0.010 | 10% | not listed |
| Mistral: Mistral Small 4 | $0.15 | $0.015 | 10% | not listed |
| Mistral: Mixtral 8x22B Instruct | $2.00 | $0.20 | 10% | not listed |
| Mistral: Saba | $0.20 | $0.020 | 10% | not listed |
| Mistral: Voxtral Small 24B 2507 | $0.10 | $0.010 | 10% | not listed |
| MoonshotAI Kimi Latest | $3.00 | $0.30 | 10% | not listed |
| MoonshotAI: Kimi K3 | $3.00 | $0.30 | 10% | not listed |
| Nex AGI: Nex-N2-Mini | $0.025 | $0.0025 | 10% | not listed |
| Nex AGI: Nex-N2-Pro | $0.25 | $0.025 | 10% | not listed |
| OpenAI GPT Latest | $5.00 | $0.50 | 10% | $6.25 |
| OpenAI GPT Mini Latest | $0.75 | $0.075 | 10% | not listed |
| OpenAI: GPT Chat Latest | $5.00 | $0.50 | 10% | not listed |
| OpenAI: GPT-5 | $1.25 | $0.13 | 10% | not listed |
| OpenAI: GPT-5 Chat | $1.25 | $0.13 | 10% | not listed |
| OpenAI: GPT-5 Codex | $1.25 | $0.13 | 10% | not listed |
| OpenAI: GPT-5 Image Mini | $2.50 | $0.25 | 10% | not listed |
| OpenAI: GPT-5 Mini | $0.25 | $0.025 | 10% | not listed |
| OpenAI: GPT-5 Nano | $0.050 | $0.0050 | 10% | not listed |
| OpenAI: GPT-5.1 | $1.25 | $0.13 | 10% | not listed |
| OpenAI: GPT-5.1 Chat | $1.25 | $0.13 | 10% | not listed |
| OpenAI: GPT-5.1-Codex | $1.25 | $0.13 | 10% | not listed |
| OpenAI: GPT-5.1-Codex-Max | $1.25 | $0.13 | 10% | not listed |
| OpenAI: GPT-5.1-Codex-Mini | $0.25 | $0.025 | 10% | not listed |
| OpenAI: GPT-5.2 | $1.75 | $0.17 | 10% | not listed |
| OpenAI: GPT-5.2 Chat | $1.75 | $0.17 | 10% | not listed |
| OpenAI: GPT-5.2-Codex | $1.75 | $0.17 | 10% | not listed |
| OpenAI: GPT-5.3 Chat | $1.75 | $0.17 | 10% | not listed |
| OpenAI: GPT-5.3-Codex | $1.75 | $0.17 | 10% | not listed |
| OpenAI: GPT-5.4 | $2.50 | $0.25 | 10% | not listed |
| OpenAI: GPT-5.4 Mini | $0.75 | $0.075 | 10% | not listed |
| OpenAI: GPT-5.4 Nano | $0.20 | $0.020 | 10% | not listed |
| OpenAI: GPT-5.5 | $5.00 | $0.50 | 10% | not listed |
| OpenAI: GPT-5.6 Luna | $1.00 | $0.10 | 10% | $1.25 |
| OpenAI: GPT-5.6 Luna Pro | $1.00 | $0.10 | 10% | $1.25 |
| OpenAI: GPT-5.6 Sol | $5.00 | $0.50 | 10% | $6.25 |
| OpenAI: GPT-5.6 Sol Pro | $5.00 | $0.50 | 10% | $6.25 |
| OpenAI: GPT-5.6 Terra | $2.50 | $0.25 | 10% | $3.13 |
| OpenAI: GPT-5.6 Terra Pro | $2.50 | $0.25 | 10% | $3.13 |
| Poolside: Laguna S 2.1 | $0.10 | $0.010 | 10% | not listed |
| Sakana: Fugu Ultra | $5.00 | $0.50 | 10% | not listed |
| Upstage: Solar Pro 3 | $0.15 | $0.015 | 10% | not listed |
| Meta: Muse Spark 1.1 | $1.25 | $0.15 | 12% | not listed |
| Anthropic: Claude 3 Haiku | $0.25 | $0.030 | 12% | $0.30 |
| OpenAI: GPT-5 Image | $10.00 | $1.25 | 13% | not listed |
| xAI: Grok 4.5 | $2.00 | $0.30 | 15% | not listed |
| xAI: Grok Latest | $2.00 | $0.30 | 15% | not listed |
| xAI: Grok 4.20 | $1.25 | $0.20 | 16% | not listed |
| xAI: Grok 4.20 Multi-Agent | $1.25 | $0.20 | 16% | not listed |
| xAI: Grok 4.3 | $1.25 | $0.20 | 16% | not listed |
| MoonshotAI: Kimi K2.5 | $0.57 | $0.095 | 17% | not listed |
| Thinking Machines: Inkling | $1.00 | $0.17 | 17% | not listed |
| Z.ai: GLM 4.5 | $0.60 | $0.11 | 18% | not listed |
| Z.ai: GLM 4.5V | $0.60 | $0.11 | 18% | not listed |
| Z.ai: GLM 4.6V | $0.30 | $0.055 | 18% | not listed |
| Z.ai: GLM 5.1 | $0.97 | $0.18 | 19% | not listed |
| Z.ai: GLM 5.2 | $0.82 | $0.15 | 19% | not listed |
| Z.ai: GLM 4.5 Air | $0.13 | $0.025 | 19% | not listed |
| MoonshotAI: Kimi K2.7 Code | $0.82 | $0.16 | 20% | not listed |
| DeepSeek: DeepSeek V4 Flash | $0.094 | $0.019 | 20% | not listed |
| inclusionAI: Ling-2.6-1T | $0.075 | $0.015 | 20% | not listed |
| inclusionAI: Ling-2.6-flash | $0.010 | $0.0020 | 20% | not listed |
| inclusionAI: Ring-2.6-1T | $0.075 | $0.015 | 20% | not listed |
| Kwaipilot: KAT-Coder-Air V2.5 | $0.15 | $0.030 | 20% | not listed |
| Kwaipilot: KAT-Coder-Pro V2 | $0.30 | $0.060 | 20% | not listed |
| MiniMax: MiniMax M2.7 | $0.25 | $0.050 | 20% | not listed |
| MiniMax: MiniMax M3 | $0.30 | $0.060 | 20% | not listed |
| Qwen: Qwen-Plus | $0.26 | $0.052 | 20% | $0.33 |
| Qwen: Qwen3 Coder Flash | $0.20 | $0.039 | 20% | $0.24 |
| Qwen: Qwen3 Coder Plus | $0.65 | $0.13 | 20% | $0.81 |
| Qwen: Qwen3 Max | $0.78 | $0.16 | 20% | $0.97 |
| Qwen: Qwen3.7 Max | $1.48 | $0.29 | 20% | $1.84 |
| Qwen: Qwen3.7 Plus | $0.32 | $0.064 | 20% | $0.40 |
| StepFun: Step 3.7 Flash | $0.20 | $0.040 | 20% | not listed |
| xAI: Grok Build 0.1 | $1.00 | $0.20 | 20% | not listed |
| Z.ai: GLM 4.6 | $0.50 | $0.10 | 20% | not listed |
| Z.ai: GLM 4.7 | $0.40 | $0.080 | 20% | not listed |
| Z.ai: GLM 5 Turbo | $1.20 | $0.24 | 20% | not listed |
| Z.ai: GLM 5V Turbo | $1.20 | $0.24 | 20% | not listed |
| Kwaipilot: KAT-Coder-Pro V2.5 | $0.74 | $0.15 | 20% | not listed |
| MoonshotAI: Kimi K2.6 | $0.68 | $0.14 | 21% | not listed |
| Z.ai: GLM 5 | $0.95 | $0.20 | 21% | not listed |
| Arcee AI: Trinity Large Thinking | $0.25 | $0.060 | 24% | not listed |
| AionLabs: Aion-2.0 | $0.80 | $0.20 | 25% | not listed |
| AionLabs: Aion-3.0 | $3.00 | $0.75 | 25% | not listed |
| Amazon: Nova Premier 1.0 | $2.50 | $0.63 | 25% | not listed |
| MoonshotAI: Kimi K2 Thinking | $0.60 | $0.15 | 25% | not listed |
| OpenAI: GPT-4.1 | $2.00 | $0.50 | 25% | not listed |
| OpenAI: GPT-4.1 Mini | $0.40 | $0.10 | 25% | not listed |
| OpenAI: GPT-4.1 Nano | $0.10 | $0.025 | 25% | not listed |
| OpenAI: GPT-5.4 Image 2 | $8.00 | $2.00 | 25% | not listed |
| OpenAI: o3 | $2.00 | $0.50 | 25% | not listed |
| OpenAI: o3 Deep Research | $10.00 | $2.50 | 25% | not listed |
| OpenAI: o4 Mini | $1.10 | $0.28 | 25% | not listed |
| OpenAI: o4 Mini Deep Research | $2.00 | $0.50 | 25% | not listed |
| OpenAI: o4 Mini High | $1.10 | $0.28 | 25% | not listed |
| Tencent: Hy3 | $0.14 | $0.035 | 25% | not listed |
| AionLabs: Aion-3.0-Mini | $0.70 | $0.18 | 26% | not listed |
| MiniMax: MiniMax M2.5 | $0.15 | $0.050 | 33% | not listed |
| NVIDIA: Nemotron 3 Ultra | $0.60 | $0.20 | 33% | not listed |
| Qwen: Qwen3 Coder 480B A35B | $0.30 | $0.10 | 33% | not listed |
| Tencent: Hy3 preview | $0.063 | $0.021 | 33% | not listed |
| TheDrummer: Skyfall 36B V2 | $0.55 | $0.25 | 45% | not listed |
| Qwen: Qwen3 VL 235B A22B Instruct | $0.21 | $0.10 | 48% | not listed |
| DeepSeek: DeepSeek V3 0324 | $0.27 | $0.14 | 50% | not listed |
| DeepSeek: DeepSeek V3.1 Terminus | $0.27 | $0.14 | 50% | not listed |
| DeepSeek: DeepSeek V3.2 | $0.27 | $0.13 | 50% | not listed |
| Meta: Llama 3.1 8B Instruct | $0.050 | $0.025 | 50% | not listed |
| OpenAI: GPT-4o | $2.50 | $1.25 | 50% | not listed |
| OpenAI: GPT-4o (2024-08-06) | $2.50 | $1.25 | 50% | not listed |
| OpenAI: GPT-4o (2024-11-20) | $2.50 | $1.25 | 50% | not listed |
| OpenAI: GPT-4o-mini | $0.15 | $0.075 | 50% | not listed |
| OpenAI: GPT-4o-mini (2024-07-18) | $0.15 | $0.075 | 50% | not listed |
| OpenAI: gpt-oss-safeguard-20b | $0.075 | $0.037 | 50% | not listed |
| OpenAI: o1 | $15.00 | $7.50 | 50% | not listed |
| OpenAI: o3 Mini | $1.10 | $0.55 | 50% | not listed |
| OpenAI: o3 Mini High | $1.10 | $0.55 | 50% | not listed |
| Poolside: Laguna M.1 | $0.20 | $0.10 | 50% | not listed |
| Poolside: Laguna XS 2.1 | $0.060 | $0.030 | 50% | not listed |
| Qwen: Qwen2.5 VL 72B Instruct | $0.80 | $0.40 | 50% | not listed |
| TheDrummer: Cydonia 24B V4.1 | $0.30 | $0.15 | 50% | not listed |
| DeepSeek: DeepSeek V3.1 | $0.25 | $0.13 | 52% | not listed |
| Qwen: Qwen3 Coder Next | $0.11 | $0.070 | 64% | not listed |
| DeepSeek: R1 0528 | $0.50 | $0.35 | 70% | not listed |
| Qwen: Qwen3 Next 80B A3B Instruct | $0.10 | $0.070 | 70% | not listed |
| ByteDance: UI-TARS 7B | $0.10 | $0.10 | 100% | not listed |
| IBM: Granite 4.1 8B | $0.050 | $0.050 | 100% | not listed |
| OpenAI: gpt-oss-20b | $0.030 | $0.030 | 100% | not listed |
Caching questions
What is prompt caching?
Providers can store a repeated prompt prefix and charge a reduced rate when a request reads it again. The cache read price is what repeated input costs; the cache write price, where listed, is what storing the prefix costs.
What does the discount column mean?
It is the listed cache read price divided by the listed input price, shown as a percentage. A 10% value means cached input costs one tenth of fresh input on OpenRouter's listing. It is a ratio of two prices from one source, not a quality signal.
Why is a model missing?
Only models OpenRouter lists with both an input price and a cache read price appear. Models without listed cache pricing are omitted rather than guessed. Prices retrieved July 22, 2026.