PRICE~deepseek/deepseek-v4-flash-latest cached_input_per_mtok increased from 0.0051 to 0.01313mPRICEdeepseek/deepseek-v4-flash-0731 input_per_mtok increased from 0.0051 to 0.01713mPRICEdeepseek/deepseek-v4-flash-0731 cached_input_per_mtok increased from 0.0051 to 0.01713mPRICEnvidia/nemotron-3-ultra-550b-a55b input_per_mtok increased from 0.50 to 0.6013mPRICEnvidia/nemotron-3-ultra-550b-a55b output_per_mtok increased from 2.20 to 2.4013mPRICEnvidia/nemotron-3-ultra-550b-a55b cached_input_per_mtok increased from 0.10 to 0.1213mPRICEtencent/hy3 input_per_mtok decreased from 0.13 to 0.08313mPRICEtencent/hy3 output_per_mtok decreased from 0.53 to 0.3313mPRICEtencent/hy3 cached_input_per_mtok decreased from 0.033 to 0.02113mPRICEtencent/hy4-preview input_per_mtok decreased from 0.83 to 0.7513mPRICEtencent/hy4-preview output_per_mtok decreased from 2.50 to 2.2513mPRICEtencent/hy4-preview cached_input_per_mtok decreased from 0.042 to 0.03813mPRICEz-ai/glm-5.3-flash input_per_mtok decreased from 0.15 to 0.02613mPRICEz-ai/glm-5.3-flash output_per_mtok increased from 0.50 to 0.9013mPRICEz-ai/glm-5.3-flash cached_input_per_mtok decreased from 0.03 to 0.01113mPRICE~deepseek/deepseek-v4-flash-latest input_per_mtok increased from 0.0051 to 0.01313mPRICE~deepseek/deepseek-v4-flash-latest output_per_mtok increased from 1.28 to 1.6013mPRICEdeepseek/deepseek-v4-flash-0731 input_per_mtok decreased from 0.012 to 0.005144mPRICEdeepseek/deepseek-v4-flash-0731 cached_input_per_mtok decreased from 0.012 to 0.005144mPRICE~deepseek/deepseek-v4-flash-latest input_per_mtok decreased from 0.0086 to 0.005144mPRICE~deepseek/deepseek-v4-flash-latest output_per_mtok decreased from 1.60 to 1.2844mPRICE~deepseek/deepseek-v4-flash-latest cached_input_per_mtok decreased from 0.0086 to 0.005144mPRICEdeepseek/deepseek-v4-flash-0731 input_per_mtok decreased from 0.017 to 0.0121hPRICEdeepseek/deepseek-v4-flash-0731 cached_input_per_mtok decreased from 0.017 to 0.0121hPRICE~deepseek/deepseek-v4-flash-latest input_per_mtok decreased from 0.013 to 0.00861hPRICE~deepseek/deepseek-v4-flash-latest cached_input_per_mtok decreased from 0.013 to 0.00861hNEW MODELinclusionai/ling-3.1-flash appeared in the catalogue2hPRICEdeepseek/deepseek-v4-flash-0731 input_per_mtok increased from 0.0051 to 0.0172hPRICEdeepseek/deepseek-v4-flash-0731 cached_input_per_mtok increased from 0.0051 to 0.0172hPRICE~deepseek/deepseek-v4-flash-latest input_per_mtok increased from 0.0038 to 0.0132hPRICE~deepseek/deepseek-v4-flash-latest cached_input_per_mtok increased from 0.0038 to 0.0132hPRICEDeepSeek V4 Flash via OpenRouter: input price ↓ 33%2hPRICEDeepSeek V4 Flash via OpenRouter: output price ↓ 33%2hPRICEdeepseek/deepseek-v4-flash input_per_mtok decreased from 0.042 to 0.0282hPRICEdeepseek/deepseek-v4-flash output_per_mtok decreased from 0.084 to 0.0562hPRICEdeepseek/deepseek-v4-flash cached_input_per_mtok decreased from 0.0084 to 0.00562hPRICEdeepseek/deepseek-v4-flash-0731 input_per_mtok decreased from 0.012 to 0.00512hPRICEdeepseek/deepseek-v4-flash-0731 cached_input_per_mtok decreased from 0.012 to 0.00512hPRICE~deepseek/deepseek-v4-flash-latest input_per_mtok decreased from 0.0086 to 0.00382hPRICE~deepseek/deepseek-v4-flash-latest cached_input_per_mtok decreased from 0.0086 to 0.00382h

OpenRouter vs Together

10 shared models · US vs US · independent benchmarks

Registry facts

FactOpenRouterTogether
Typerouterinference host
HeadquartersUSUS
Catalogued endpoints11
EU residencyNoNo

GLM-5.2: OpenRouter vs Together

● MEASURED OpenRouter detail → · ● MEASURED Together detail →

MetricOpenRouterTogether
TTFT p50480 ms271 ms
TTFT p951131 ms589 ms
Throughput137 tok/s283 tok/s
Reliability99.1%100%
Tool calling100%100%
Structured output71.4%100%
$ output / M tokens$3.99$4.40

DeepSeek V4 Flash: OpenRouter vs Together

● MEASURED OpenRouter detail → · ● MEASURED Together detail →

MetricOpenRouterTogether
TTFT p501059 ms714 ms
TTFT p958863 ms1866 ms
Throughput89 tok/s61 tok/s
Reliability100%98.6%
Tool calling100%100%
Structured output92.9%100%
$ output / M tokens$0.056$0.28

Kimi K3: OpenRouter vs Together

● MEASURED OpenRouter detail → · ● MEASURED Together detail →

MetricOpenRouterTogether
TTFT p50674 ms602 ms
TTFT p953408 ms1514 ms
Throughput90 tok/s104 tok/s
Reliability97.2%99.5%
Tool calling100%100%
Structured output100%97.6%
$ output / M tokens$13.50$13.50

MiniMax M3: OpenRouter vs Together

● MEASURED OpenRouter detail → · ● MEASURED Together detail →

MetricOpenRouterTogether
TTFT p50929 ms542 ms
TTFT p953033 ms1736 ms
Throughput75 tok/s132 tok/s
Reliability99.5%100%
Tool calling100%100%
Structured output97.6%97.6%
$ output / M tokens$1.20$1.20

GPT-OSS 120B: OpenRouter vs Together

● MEASURED OpenRouter detail → · ● MEASURED Together detail →

MetricOpenRouterTogether
TTFT p50474 ms182 ms
TTFT p951413 ms467 ms
Throughput68 tok/s138 tok/s
Reliability99.1%93.5%
Tool calling100%100%
Structured output100%85.7%
$ output / M tokens$0.17$0.60

Llama 3.3 70B: OpenRouter vs Together

● MEASURED OpenRouter detail → · ● MEASURED Together detail →

MetricOpenRouterTogether
TTFT p50576 ms632 ms
TTFT p954604 ms1801 ms
Throughput38 tok/s60 tok/s
Reliability100%100%
Tool calling100%98%
Structured output100%100%
$ output / M tokens$0.32$1.04

Every figure above is an independent measurement from the canonical vantage; unmeasured fields show — until measured. How we benchmark →