Z.ai: GLM 5.3 Flash
Context 1.0Mz-ai/glm-5.3-flash
GLM-5.3-Flash является моделью Z.ai с изначально встроенной мультимодальностью. Она подходит для эффективного программирования и длительных задач ИИ-агентов. Гибридная архитектура, сочетающая разреженное и линейное внимание, сохраняет точность работы с длинным контекстом и одновременно снижает вычислительные затраты.
Возможности
Chat Completions
Streaming
Vision
Files
Вход
Изображения · Текст · Видео
Выход
Текст
Поддерживаемые параметры API
frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_pparallel_tool_callspresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p
Лимиты
Maximum output943.7K
Параметры по умолчанию
{
"top_p": 0.95,
"temperature": 1
}Провайдеры
| Провайдер | Cache read/1M | Context | |||||
|---|---|---|---|---|---|---|---|
Relace relace | 3.77 ₽ | 6.96 ₽ | 2.71 ₽ | 1.0M | 0.76 s | 89 tps | 100.00% |
Wafer wafer/nvfp4 | 3.98 ₽ | 18.08 ₽ | 3.89 ₽ | 1.0M | 0.36 s | 153 tps | 100.00% |
Darkbloom darkbloom/fp4 | 5.88 ₽ | 16.27 ₽ | 2.94 ₽ | 262.1K | 2.34 s | 14 tps | — |
OpenInference open-inference/fp4 | 6.01 ₽ | 6.31 ₽ | 1.62 ₽ | 1.0M | 0.57 s | 119 tps | 100.00% |
StreamLake streamlake/fp8 | 6.78 ₽ | 22.60 ₽ | 1.36 ₽ | 1.0M | 2.25 s | 37 tps | 100.00% |
GMICloud gmicloud/fp8 | 6.78 ₽ | 22.60 ₽ | 1.36 ₽ | 1.0M | 1.35 s | 53 tps | 100.00% |
Sail Research sail-research/us | 7.09 ₽ | 22.60 ₽ | 2.36 ₽ | 1.0M | 1.23 s | 24 tps | 100.00% |
Sail Research sail-research/fp4 | 7.09 ₽ | 22.60 ₽ | 2.36 ₽ | 1.0M | 1.45 s | 22 tps | 100.00% |
Near AI near-ai/fp8 | 7.23 ₽ | 45.20 ₽ | 1.08 ₽ | 1.0M | 2.21 s | 10 tps | 100.00% |
Reka reka | 7.23 ₽ | 216.96 ₽ | 3.62 ₽ | 1.0M | 0.90 s | 31 tps | 100.00% |
Novita novita/fp8 | 7.59 ₽ | 25.31 ₽ | 1.51 ₽ | 1.0M | 4.10 s | 45 tps | 100.00% |
InferenceNet inference-net/us | 10.85 ₽ | 35.08 ₽ | 2.89 ₽ | 1.0M | 0.64 s | 118 tps | 100.00% |
InferenceNet inference-net/fp4 | 10.85 ₽ | 35.08 ₽ | 2.89 ₽ | 1.0M | 0.51 s | 123 tps | 100.00% |
NextBit nextbit/fp4 | 11.30 ₽ | 45.20 ₽ | 4.52 ₽ | 1.0M | 2.53 s | 20 tps | 100.00% |
Phala phala/nvfp4 | 11.53 ₽ | 38.42 ₽ | 2.31 ₽ | 1.0M | 3.97 s | 26 tps | 100.00% |
DigitalOcean digitalocean | 11.87 ₽ | 39.55 ₽ | 2.71 ₽ | 1.0M | 0.95 s | 108 tps | 100.00% |
Makora makora | 12.66 ₽ | 42.49 ₽ | 2.17 ₽ | 1.0M | 0.21 s | 21 tps | 100.00% |
Mistral mistral | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 1.32 s | 103 tps | 100.00% |
Cohere cohere | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 0.81 s | 147 tps | 100.00% |
PrimeIntellect primeintellect | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 0.87 s | 41 tps | 100.00% |
Nebius nebius/fp4 | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 3.10 s | 37 tps | 57.14% |
Modal modal/nvfp4 | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | — | — | — |
BaseTen baseten/fp8/us | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 0.59 s | 120 tps | — |
Crusoe crusoe/fp4 | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 0.91 s | 136 tps | 100.00% |
CoreWeave coreweave/nvfp4 | 13.56 ₽ | 45.20 ₽ | 1.81 ₽ | 1.0M | 1.59 s | 113 tps | 100.00% |
AtlasCloud atlas-cloud/fp8 | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 2.53 s | 56 tps | 100.00% |
Fireworks fireworks | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 0.85 s | 106 tps | 100.00% |
Friendli friendli | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 0.65 s | 98 tps | 100.00% |
SiliconFlow siliconflow/fp8 | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 2.24 s | 30 tps | 96.67% |
Together together | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 0.77 s | 121 tps | 100.00% |
Parasail parasail/fp4 | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 0.64 s | 103 tps | 100.00% |
BaseTen baseten/fp8 | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 0.70 s | 100 tps | 100.00% |
Venice venice | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 1.28 s | 106 tps | 100.00% |
DeepInfra deepinfra/fp4 | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 1.20 s | 17 tps | 100.00% |
Z.AI z-ai/fp8 | 13.56 ₽ | 45.20 ₽ | 2.71 ₽ | 1.0M | 4.27 s | 36 tps | 100.00% |
Inceptron inceptron/fp8 | 18.08 ₽ | 67.80 ₽ | 6.78 ₽ | 1.0M | 0.70 s | 50 tps | 100.00% |
Fireworks fireworks/us | 20.34 ₽ | 67.80 ₽ | 4.07 ₽ | 1.0M | 0.80 s | 82 tps | 100.00% |
Morph morph/fp8 | 29.38 ₽ | 58.76 ₽ | 8.23 ₽ | 1.0M | 1.47 s | 35 tps | 100.00% |