Qwen2.5 VL Instruct by Alibaba is available through 59 API providers on LMSpeed. Compare API pricing from $0.014 to $75.00 per million input tokens across providers. Context window: 131,072. In speed benchmarks, the fastest provider reaches 86 tok/s.
Alibaba Qwen2.5 VL Instruct is a vision-language model in the Qwen2.5 series, supporting image understanding alongside text generation and instruction following.
Также известна как
Compare Qwen2.5 VL Instruct API pricing across 59 providers. Input prices range from $0.014 to $75.00 per million input. ModelPool offers the lowest rate at $0.014/M.
| Провайдер | Вариант модели | Аудит | Входные данные ($/M) | Выходные данные ($/M) | Скорость (т/с) | Первый токен |
|---|---|---|---|---|---|---|
V-API100% | Qwen2.5-VL-32B-Instruct | — | $0.020 | $0.020 | — | — |
| qwen2.5-vl-72b-instruct | — | $16.00 | $48.00 | — | — | |
柏拉图AI100% | qwen2.5-vl-3b-instruct | — | $0.027 | $0.082 | — | — |
| qwen2.5-vl-7b-instruct | — | $0.041 | $0.164 | — | — | |
| qwen2.5-vl-72b-instruct | — | $0.411 | $1.64 | — | — | |
| qwen2.5-vl-32b-instruct | — | $1.10 | $3.29 | — | — | |
AAAI100% | Qwen2.5-VL-32B-Instruct | — | $0.020 | $0.020 | — | — |
| qwen2.5-vl-72b-instruct | — | $16.00 | $48.00 | — | — | |
Seamee API100% | Qwen/Qwen2.5-VL-32B-Instruct | — | $0.050 | $0.220 | — | — |
| Qwen/Qwen2.5-VL-7B-Instruct | — | $0.050 | $0.050 | — | — | |
| Qwen/Qwen2.5-VL-72B-Instruct | — | $0.250 | $0.750 | — | — | |
| Qwen/Qwen2.5-VL-3B-Instruct | — | $2.00 | $2.00 | — | — | |
毫秒API100% | qwen2.5-vl-3b-instruct | — | $0.200 | $0.600 | — | — |
| qwen2.5-vl-7b-instruct | — | $0.300 | $1.20 | — | — | |
| qwen2.5-vl-72b-instruct | — | $3.00 | $12.00 | — | — | |
| qwen2.5-vl-32b-instruct | — | $8.00 | $24.00 | — | — | |
Newagiai100% | qwen2.5-vl-3b-instruct | — | $0.027 | $0.082 | — | — |
| qwen2.5-vl-7b-instruct | — | $0.041 | $0.164 | — | — | |
| qwen2.5-vl-72b-instruct | — | $0.411 | $1.64 | — | — | |
| qwen2.5-vl-32b-instruct | — | $1.10 | $3.29 | — | — |
Данные о ценах из публичных API провайдеров
Сравните производительность по скорости и задержке у всех API-провайдеров.
| Провайдер | Скорость | Задержка | Тесты |
|---|---|---|---|
联无所AI Pro/Qwen/Qwen2.5-VL-7B-Instruct | 85.51 tok/s | 0.76s | 5 |
qwen2.5-vl-72b-instruct | 43.29 tok/s | 1.03s | 5 |
Qwen/Qwen2.5-VL-32B-Instruct | 30.07 tok/s | 0.87s | 5 |
Qwen/Qwen2.5-VL-72B-Instruct | 28.12 tok/s | 0.92s | 10 |
Показано 1–4 из 4 провайдеров
deepseek-v3
DeepSeek V3 is DeepSeek flagship MoE language model with 671B total parameters, delivering strong performance in reasoning, coding, and multilingual tasks at competitive inference cost.
deepseek-v3-2
DeepSeek V3.2 is a large language model in the DeepSeek V3 series, offering advanced reasoning, code generation, and multimodal capabilities.
gemini-2-5-pro
Google Gemini 2.5 Pro is Google advanced multimodal model with a 1M-token context window, strong STEM reasoning, and native support for images, audio, and video understanding.
glm-4-7
Zhipu GLM-4.7 is a flagship GLM release from Zhipu AI with advanced Chinese-English reasoning, coding, and agent features.
qwen3
Alibaba Qwen3 is the Qwen family's flagship LLM series with dense and MoE variants, seamless thinking/non-thinking modes, and leading open-source performance in math, code, and agent tasks.
claude-opus-4-6
Anthropic Claude Opus 4.6 is the most capable Claude Opus tier, optimized for complex analysis, long-horizon coding, and high-stakes enterprise reasoning workloads.
Рейтинги основаны на тестах, предоставленных сообществом, и периодических зондах работоспособности. Носит рекомендательный характер, не является официальными данными.
