API pricing
Price changes87 models, cheapest first by blended cost per million tokens. Every price is a LiteLLM consensus number, independently cross-checked against other aggregators — disputed means the cross-check disagrees.
| Model | Developer | Input $/M | Output $/M | Blended (3:1) | Context | Verification |
|---|---|---|---|---|---|---|
| GLM-4.7-Flash | Zhipu AI | free | free | free | 200,000 | disputed |
| GPT-5 nano | OpenAI | $0.05 | $0.4 | $0.138 | 272,000 | disputed |
| Mistral NeMo | Mistral AI | $0.15 | $0.15 | $0.15 | 128,000 | disputed |
| GPT-4.1 nano | OpenAI | $0.1 | $0.4 | $0.175 | 1,047,576 | corroborated |
| GPT-4o mini | OpenAI | $0.15 | $0.6 | $0.262 | 128,000 | corroborated |
| Grok 4.1 Fast | xAI | $0.2 | $0.5 | $0.275 | 2,000,000 | single-source |
| Qwen3-235B-A22B | Qwen | $0.2 | $0.6 | $0.3 | 262,144 | disputed |
| Qwen2.5-7B-Instruct | Qwen | $0.175 | $0.7 | $0.306 | 131,072 | disputed |
| Qwen3-8B | Qwen | $0.18 | $0.7 | $0.31 | 131,072 | disputed |
| DeepSeek-V3.2 | DeepSeek | $0.28 | $0.4 | $0.31 | 163,840 | corroborated |
| DeepSeek-R1 | DeepSeek | $0.28 | $0.42 | $0.315 | 131,072 | disputed |
| Grok-3 mini | xAI | $0.3 | $0.5 | $0.35 | 131,072 | single-source |
| Qwen 3.6 Flash | Qwen | $0.188 | $1.13 | $0.422 | 1,000,000 | corroborated |
| GPT-5.6 Luna | OpenAI | $0.2 | $1.2 | $0.45 | 922,000 | single-source |
| GPT-5.4 Nano | OpenAI | $0.2 | $1.25 | $0.463 | 272,000 | disputed |
| DeepSeek-V3 | DeepSeek | $0.27 | $1.1 | $0.478 | 65,536 | single-source |
| Claude 3 Haiku | Anthropic | $0.25 | $1.25 | $0.5 | 200,000 | corroborated |
| MiniMax-M2.5 | MiniMax | $0.3 | $1.2 | $0.525 | 1,000,000 | disputed |
| Gemini 3.1 Flash-Lite | Google DeepMind | $0.25 | $1.5 | $0.563 | 1,048,576 | corroborated |
| Qwen2.5-14B-Instruct | Qwen | $0.35 | $1.4 | $0.612 | 131,072 | single-source |
| Qwen3-14B | Qwen | $0.35 | $1.4 | $0.612 | 131,072 | disputed |
| DeepSeek-V4-Flash | DeepSeek | $0.44 | $1.32 | $0.66 | 1,000,000 | disputed |
| Qwen3-Coder-Next | Qwen | $0.5 | $1.2 | $0.675 | 262,144 | disputed |
| GPT-5 mini | OpenAI | $0.25 | $2 | $0.688 | 272,000 | disputed |
| GPT-4.1 mini | OpenAI | $0.4 | $1.6 | $0.7 | 1,047,576 | corroborated |
| Mistral Large 3 | Mistral AI | $0.5 | $1.5 | $0.75 | 262,144 | single-source |
| Mistral Medium 3 | Mistral AI | $0.4 | $2 | $0.8 | 128,000 | disputed |
| Qwen3-Coder-30B-A3B-Instruct | Qwen | $0.45 | $2.25 | $0.9 | 262,144 | disputed |
| GLM-4.6 | Zhipu AI | $0.6 | $2.2 | $1 | 200,000 | disputed |
| GLM-4.7 | Zhipu AI | $0.6 | $2.2 | $1 | 200,000 | disputed |
| Kimi K2 Thinking | Moonshot AI | $0.6 | $2.5 | $1.07 | 262,144 | corroborated |
| Gemini 3 Flash | Google DeepMind | $0.5 | $3 | $1.13 | 1,048,576 | single-source |
| Qwen 3.6 Plus | Qwen | $0.5 | $3 | $1.13 | 1,000,000 | disputed |
| Kimi K2.5 | Moonshot AI | $0.6 | $3 | $1.2 | 262,144 | disputed |
| Qwen3-32B | Qwen | $0.7 | $2.8 | $1.22 | 131,072 | disputed |
| Qwen2.5-32B-Instruct | Qwen | $0.7 | $2.8 | $1.22 | 131,072 | single-source |
| Amazon Nova Pro | Amazon | $0.8 | $3.2 | $1.4 | 300,000 | single-source |
| GLM-5 | Zhipu AI | $1 | $3.2 | $1.55 | 200,000 | disputed |
| GPT-5.4 Mini | OpenAI | $0.75 | $4.5 | $1.69 | 272,000 | disputed |
| Kimi K2.6 | NVIDIA | $0.95 | $4 | $1.71 | 262,144 | disputed |
| Kimi K2.7 Code | Moonshot AI | $0.95 | $4 | $1.71 | 262,144 | disputed |
| o3-mini | OpenAI | $1.1 | $4.4 | $1.93 | 200,000 | corroborated |
| o4-mini | OpenAI | $1.1 | $4.4 | $1.93 | 200,000 | corroborated |
| DeepSeek-V4-Pro | DeepSeek | $1.32 | $3.96 | $1.98 | 1,000,000 | disputed |
| Claude Haiku 4.5 | Anthropic | $1 | $5 | $2 | 200,000 | corroborated |
| GLM-5.2 | Zhipu AI | $1.4 | $4.4 | $2.15 | 1,000,000 | disputed |
| GLM-5.1 | Zhipu AI | $1.4 | $4.4 | $2.15 | 200,000 | disputed |
| Qwen2.5-72B-Instruct | Qwen | $1.4 | $5.6 | $2.45 | 131,072 | disputed |
| Qwen 3.6 Max (Preview) | Qwen | $1.3 | $7.8 | $2.92 | 262,144 | disputed |
| Gemini 3.5 Flash | Google DeepMind | $1.5 | $9 | $3.38 | 1,048,576 | corroborated |
| GPT-5 | OpenAI | $1.25 | $10 | $3.44 | 272,000 | disputed |
| GPT-5.1 | OpenAI | $1.25 | $10 | $3.44 | 272,000 | disputed |
| o3 | OpenAI | $2 | $8 | $3.5 | 200,000 | corroborated |
| GPT-4.1 | OpenAI | $2 | $8 | $3.5 | 1,047,576 | corroborated |
| Qwen3-Max | Qwen | $2.11 | $8.45 | $3.69 | 258,048 | disputed |
| Qwen3.7-Max | Qwen | $2.5 | $7.5 | $3.75 | 1,000,000 | disputed |
| Claude Sonnet 5 | Anthropic | $2 | $10 | $4 | 1,000,000 | single-source |
| GPT-4o | OpenAI | $2.5 | $10 | $4.38 | 128,000 | single-source |
| Gemini 3 Pro | Google DeepMind | $2 | $12 | $4.5 | 1,048,576 | single-source |
| Gemini 3.1 Pro | Google DeepMind | $2 | $12 | $4.5 | 1,048,576 | single-source |
| GPT-5.6 Terra | OpenAI | $2 | $12 | $4.5 | 922,000 | single-source |
| Mistral Large | Mistral AI | $3 | $10 | $4.75 | 131,072 | disputed |
| GPT-5.2 | OpenAI | $1.75 | $14 | $4.81 | 272,000 | disputed |
| GPT-5.3 Codex | OpenAI | $1.75 | $14 | $4.81 | 272,000 | disputed |
| GPT-5.4 | OpenAI | $2.5 | $15 | $5.63 | 1,050,000 | corroborated |
| Grok 3 | xAI | $3 | $15 | $6 | 131,072 | single-source |
| Grok 4 | xAI | $3 | $15 | $6 | 128,000 | single-source |
| Claude 3.7 Sonnet | Anthropic | $3 | $15 | $6 | 200,000 | single-source |
| Claude Sonnet 4 | Anthropic | $3 | $15 | $6 | 1,000,000 | corroborated |
| Claude Sonnet 4.5 | Anthropic | $3 | $15 | $6 | 200,000 | disputed |
| Claude Sonnet 4.6 | Anthropic | $3 | $15 | $6 | 1,000,000 | corroborated |
| Amazon Nova 2 Pro | Amazon | $2.19 | $17.5 | $6.02 | 1,000,000 | single-source |
| Claude Opus 4.5 | Anthropic | $5 | $25 | $10 | 200,000 | corroborated |
| Claude Opus 4.6 | Anthropic | $5 | $25 | $10 | 1,000,000 | corroborated |
| Claude Opus 4.7 | Anthropic | $5 | $25 | $10 | 1,000,000 | corroborated |
| Claude Opus 4.8 | Anthropic | $5 | $25 | $10 | 1,000,000 | corroborated |
| GPT-5.5 | OpenAI | $5 | $30 | $11.25 | 1,050,000 | corroborated |
| GPT-5.6 Sol | OpenAI | $5 | $30 | $11.25 | 922,000 | single-source |
| Claude Fable 5 | Anthropic | $10 | $50 | $20 | 1,000,000 | corroborated |
| o1 | OpenAI | $15 | $60 | $26.25 | 200,000 | corroborated |
| Claude Opus 4 | Anthropic | $15 | $75 | $30 | 200,000 | corroborated |
| Claude Opus 4.1 | Anthropic | $15 | $75 | $30 | 200,000 | corroborated |
| o3-pro | OpenAI | $20 | $80 | $35 | 200,000 | corroborated |
| GPT-5 Pro | OpenAI | $15 | $120 | $41.25 | 400,000 | corroborated |
| GPT-5.2 Pro | OpenAI | $21 | $168 | $57.75 | 272,000 | disputed |
| GPT-5.4 Pro | OpenAI | $30 | $180 | $67.5 | 1,050,000 | corroborated |
| GPT-5.5 Pro | OpenAI | $30 | $180 | $67.5 | 1,050,000 | corroborated |
Every price is a LiteLLM consensus number, independently cross-checked against other pricing aggregators — never republished from a single source we haven’t verified. Blended (3:1) = (3 × input + output) / 4, the standard API-cost convention weighted toward input (most traffic is input-heavy: long context, short completions). Disputed means our cross-check source disagrees with the LiteLLM number — treat that price as directional, not exact. Every time a price, context window, or deprecation date actually moves, it lands on the append-only changes ledger →. Full methodology →