Which flagship models give the most tokens for your dollar?
Auto-updated ranking of flagship AI models by bang-for-buck: OpenRouter list pricing, context window, max output tokens and tokens-per-dollar value score.
| Rank | Model | Value | Tokens/$ | Input $/1M | Output $/1M | Blended $/1M | Context | Max output | Price move | Rank move | Source |
|---|---|---|---|---|---|---|---|---|---|---|---|
| #1 | Qwen: Qwen3.7 FlashQwen | 100 | 18.2M | $0.03 | $0.13 | $0.055 | 1.0M | 66k | NEW | NEW | OpenRouter |
| #2 | DeepSeek: DeepSeek V4 Flash 0731Deepseek | 51.2 | 8.9M | $0.09 | $0.18 | $0.1125 | 1.0M | 384k | NEW | NEW | OpenRouter |
| #3 | DeepSeek: DeepSeek V4 Flash 0423Deepseek | 33 | 5.7M | $0.14 | $0.28 | $0.175 | 1.0M | 393k | NEW | NEW | OpenRouter |
| #4 | OpenAI: GPT-5.6 Luna ProOpenai | 24.9 | 4.4M | $0.1 | $0.6 | $0.225 | 1.1M | 128k | NEW | NEW | OpenRouter |
| #5 | OpenAI: GPT-5.6 Luna Pro (batch)Openai | 24.9 | 4.4M | $0.1 | $0.6 | $0.225 | 1.1M | 128k | NEW | NEW | OpenRouter |
| #6 | OpenAI: GPT-5.6 LunaOpenai | 24.9 | 4.4M | $0.1 | $0.6 | $0.225 | 1.1M | 128k | NEW | NEW | OpenRouter |
| #7 | OpenAI: GPT-5.6 Luna (batch)Openai | 24.9 | 4.4M | $0.1 | $0.6 | $0.225 | 1.1M | 128k | NEW | NEW | OpenRouter |
| #8 | Z.ai: GLM 5.2Z Ai | 17.7 | 3.2M | $0.2058 | $0.6468 | $0.3161 | 1.0M | 128k | NEW | NEW | OpenRouter |
| #9 | Google: Gemini 3.5 Flash Lite (batch)Google | 13 | 2.4M | $0.15 | $1.25 | $0.425 | 1.0M | 66k | NEW | NEW | OpenRouter |
| #10 | DeepSeek: DeepSeek V4 ProDeepseek | 10.6 | 1.8M | $0.435 | $0.87 | $0.5437 | 1.0M | 384k | NEW | NEW | OpenRouter |
| #11 | Qwen: Qwen3.7 PlusQwen | 10 | 1.8M | $0.32 | $1.28 | $0.56 | 1.0M | 131k | NEW | NEW | OpenRouter |
| #12 | NVIDIA: Nemotron 3 Ultra (batch)Nvidia | 7.4 | 1.5M | $0.3 | $1.8 | $0.675 | 512k | — | NEW | NEW | OpenRouter |
| #13 | Google: Gemini 3.5 Flash LiteGoogle | 6.5 | 1.2M | $0.3 | $2.5 | $0.85 | 1.0M | 66k | NEW | NEW | OpenRouter |
| #14 | Z.ai: GLM 5.2 (batch)Z Ai | 4.7 | 930k | $0.7 | $2.2 | $1.075 | 512k | — | NEW | NEW | OpenRouter |
| #15 | SpaceXAI: Grok Build 0.1X Ai | 3.9 | 800k | $1 | $2 | $1.25 | 256k | — | NEW | NEW | OpenRouter |
| #16 | MoonshotAI: Kimi K2.7 CodeMoonshotai | 3.8 | 714k | $0.7 | $3.5 | $1.4 | 262k | 262k | NEW | NEW | OpenRouter |
| #17 | NVIDIA: Nemotron 3 UltraNvidia | 3.7 | 741k | $0.6 | $3.6 | $1.35 | 512k | — | NEW | NEW | OpenRouter |
| #18 | Google: Gemini 3.6 Flash (batch)Google | 3.7 | 667k | $0.75 | $3.75 | $1.5 | 1.0M | 66k | NEW | NEW | OpenRouter |
| #19 | SpaceXAI: Grok 4.3X Ai | 3.3 | 640k | $1.25 | $2.5 | $1.5625 | 1.0M | — | NEW | NEW | OpenRouter |
| #20 | Qwen: Qwen3.8 MaxQwen | 1.9 | 333k | $2 | $6 | $3 | 1.0M | 131k | NEW | NEW | OpenRouter |
| #21 | Google: Gemini 3.6 FlashGoogle | 1.8 | 333k | $1.5 | $7.5 | $3 | 1.0M | 66k | NEW | NEW | OpenRouter |
| #22 | SpaceXAI: Grok 4.5X Ai | 1.7 | 333k | $2 | $6 | $3 | 500k | — | NEW | NEW | OpenRouter |
| #23 | Mistral: Mistral Medium 3.5Mistralai | 1.6 | 333k | $1.5 | $7.5 | $3 | 262k | — | NEW | NEW | OpenRouter |
| #24 | Anthropic: Claude Sonnet 5Anthropic | 1.4 | 250k | $2 | $10 | $4 | 1.0M | 128k | NEW | NEW | OpenRouter |
| #25 | Claude Opus 5 (batch)Anthropic | 1.1 | 200k | $2.5 | $12.5 | $5 | 1.0M | 128k | NEW | NEW | OpenRouter |
| #26 | MoonshotAI: Kimi K3Moonshotai | 0.9 | 167k | $3 | $15 | $6 | 1.0M | — | NEW | NEW | OpenRouter |
| #27 | Claude Opus 5Anthropic | 0.6 | 100k | $5 | $25 | $10 | 1.0M | 128k | NEW | NEW | OpenRouter |
| #28 | Claude Opus 5 (Fast)Anthropic | 0.3 | 50k | $10 | $50 | $20 | 1.0M | 128k | NEW | NEW | OpenRouter |
Dynamic flagship selection picks the newest paid text models from major providers on OpenRouter. Blended price uses a 3:1 input-to-output token mix in USD per million tokens. Tokens per dollar is 1,000,000 divided by that blended price. Value score (0-100) combines tokens-per-dollar with a capability factor from context window and max output tokens (log-scaled). It ranks list-price efficiency, not benchmark quality. This ranks list-price efficiency from OpenRouter Models API, not independent benchmark quality. Check provider pages for latency, safety filters and enterprise terms before production use.