大模型 API 价格变动时间线(降价/调价记录)· 大模型库

每日自动比对各模型价格,有变动即记录在此。绿色为降价、红色为涨价; 点击模型名查看该模型完整价格历史曲线。← 返回总表

2026-08-27

Qwen: Qwen3 235B A22B Instruct 2507最大输出16K → 236K
Qwen: Qwen3 235B A22B Instruct 2507缓存读— → $0.02
Qwen: Qwen3 235B A22B Instruct 2507输出价$0.55 → $0.35-36%
Qwen: Qwen3 235B A22B Instruct 2507输入价$0.09 → $0.09-3%
Z.ai: GLM 4.6最大输出131K → 16K
Z.ai: GLM 4.6缓存读$0.10 → $0.08-20%
Z.ai: GLM 4.6输出价$2.00 → $1.75-12%
Z.ai: GLM 4.6输入价$0.50 → $0.43-14%
NVIDIA: Nemotron 3 Nano 30B A3B最大输出236K → 228K
NVIDIA: Nemotron 3 Nano 30B A3B缓存读$0.03 → $0.03-17%
Qwen: Qwen3.5 397B A17B缓存读$0.30 → —
DeepSeek: DeepSeek V4 Flash 0423缓存读$0.02 → $0.01-4%
DeepSeek: DeepSeek V4 Flash 0423输出价$0.16 → $0.15-4%
DeepSeek: DeepSeek V4 Flash 0423输入价$0.08 → $0.07-4%
Qwen: Qwen3.6 27B最大输出82K → 236K
Qwen: Qwen3.6 27B缓存读— → $0.12
Qwen: Qwen3.6 27B输出价$3.20 → $3.60+12%
Qwen: Qwen3.6 27B输入价$0.32 → $0.60+87%
Qwen: Qwen3.6 35B A3B输出价$1.00 → $0.90-10%
Qwen: Qwen3.6 35B A3B输入价$0.14 → $0.10-29%
NVIDIA: Nemotron 3 Ultra最大输出461K → 16K
NVIDIA: Nemotron 3 Ultra上下文窗口512K → 262K
NVIDIA: Nemotron 3 Ultra缓存读$0.20 → $0.10-50%
NVIDIA: Nemotron 3 Ultra输出价$3.60 → $2.20-39%
NVIDIA: Nemotron 3 Ultra输入价$0.60 → $0.50-17%
MoonshotAI: Kimi K2.7 Code缓存读$0.19 → $0.18-5%
MoonshotAI: Kimi K2.7 Code输入价$0.67 → $0.66-1%
DeepSeek: DeepSeek V4 Flash 0731缓存读$0.01 → $0.01-17%
DeepSeek: DeepSeek V4 Flash 0731输出价$0.12 → $0.10-17%
DeepSeek: DeepSeek V4 Flash 0731输入价$0.06 → $0.05-17%
Qwen: Qwen3.8 Flash输入价$0.16 → $0.15-6%

2026-08-26

Meta: Llama 3.3 70B Instruct输入价$0.10 → $0.71+610%
Qwen: Qwen2.5 VL 72B Instruct缓存读$0.40 → —
Qwen: Qwen3.5 397B A17B最大输出236K → 66K
Qwen: Qwen3.5 397B A17B输出价$3.60 → $2.34-35%
Qwen: Qwen3.5 397B A17B输入价$0.50 → $0.39-22%
Google: Gemma 4 31B最大输出236K → 16K
Google: Gemma 4 31B缓存读$0.10 → $0.05-50%
Google: Gemma 4 31B输入价$0.10 → $0.09-10%
DeepSeek: DeepSeek V4 Flash 0423缓存读$0.02 → $0.02-12%
DeepSeek: DeepSeek V4 Flash 0423输出价$0.18 → $0.16-12%
DeepSeek: DeepSeek V4 Flash 0423输入价$0.09 → $0.08-12%
DeepSeek: DeepSeek V4 Pro 0423缓存读$0.07 → $0.07+10%
DeepSeek: DeepSeek V4 Pro 0423输出价$1.58 → $1.74+10%
DeepSeek: DeepSeek V4 Pro 0423输入价$0.79 → $0.87+10%
Tencent: Hy3缓存读$0.03 → $0.02-38%
Tencent: Hy3输出价$0.53 → $0.33-37%
Tencent: Hy3输入价$0.13 → $0.08-37%
DeepSeek: DeepSeek V4 Flash 0731缓存读$0.01 → $0.01+50%
DeepSeek: DeepSeek V4 Flash 0731输出价$0.08 → $0.12+50%
DeepSeek: DeepSeek V4 Flash 0731输入价$0.04 → $0.06+50%

2026-08-25

Meta: Llama 3.3 70B Instruct最大输出16K → 115K
Meta: Llama 3.3 70B Instruct缓存读— → $0.71
Meta: Llama 3.3 70B Instruct输出价$0.32 → $0.71+122%
Qwen: Qwen2.5 VL 72B Instruct最大输出115K → 29K
Qwen: Qwen2.5 VL 72B Instruct输出价$1.00 → $0.75-25%
Qwen: Qwen2.5 VL 72B Instruct输入价$0.80 → $0.25-69%
Meta: Llama 4 Scout最大输出16K → 8K
Meta: Llama 4 Scout缓存读— → $0.06
Meta: Llama 4 Scout输出价$0.30 → $0.34+13%
Meta: Llama 4 Scout输入价$0.10 → $0.11+10%
DeepSeek: DeepSeek V4 Flash 0423缓存读$0.02 → $0.02+15%
DeepSeek: DeepSeek V4 Flash 0423输出价$0.15 → $0.18+15%
DeepSeek: DeepSeek V4 Flash 0423输入价$0.08 → $0.09+15%
DeepSeek: DeepSeek V4 Pro 0423缓存读$0.05 → $0.07+42%
DeepSeek: DeepSeek V4 Pro 0423输出价$1.11 → $1.58+42%
DeepSeek: DeepSeek V4 Pro 0423输入价$0.56 → $0.79+42%
MythoMax 13B最大输出4K → 4K
OpenAI: GPT-3.5 Turbo Instruct最大输出4K → 4K
OpenAI: GPT-3.5 Turbo (older v0613)最大输出4K → 4K
Mistral Large最大输出— → 102K
Mistral: Mixtral 8x22B Instruct最大输出— → 52K
Meta: Llama 3.1 8B Instruct最大输出131K → 118K
Sao10K: Llama 3 8B Lunaris最大输出16K → 7K
Meta: Llama 3.2 3B Instruct最大输出131K → 118K
Meta: Llama 3.2 1B Instruct最大输出60K → 54K
TheDrummer: Rocinante 12B最大输出66K → 59K
Qwen: Qwen2.5 7B Instruct最大输出33K → 29K
Mistral: Ministral 8B最大输出— → 102K
TheDrummer: UnslopNemo 12B最大输出33K → 26K
Qwen2.5 Coder 32B Instruct最大输出33K → 29K
Mistral Large 2407最大输出— → 105K
Microsoft: Phi 4最大输出16K → 15K
MiniMax: MiniMax-01最大输出1M → 900K
DeepSeek: R1 Distill Llama 70B最大输出8K → 7K
Perplexity: Sonar最大输出— → 114K
Qwen: Qwen2.5 VL 72B Instruct最大输出128K → 115K
AionLabs: Aion-RP 1.0 (8B)最大输出33K → 29K
Mistral: Saba最大输出— → 26K
Perplexity: Sonar Deep Research最大输出— → 115K
Perplexity: Sonar Reasoning Pro最大输出— → 115K
TheDrummer: Skyfall 36B V2最大输出33K → 29K
Google: Gemma 3 27B最大输出131K → 118K
Reka Flash 3最大输出66K → 59K
Mistral: Mistral Small 3.1 24B最大输出128K → 102K
DeepSeek: DeepSeek V3 0324最大输出164K → 147K
Meta: Llama Guard 4 12B上下文窗口1.05M → 164K
Mistral: Mistral Medium 3最大输出— → 105K
Tencent: Hunyuan A13B Instruct最大输出131K → 118K
Qwen: Qwen3 235B A22B Thinking 2507最大输出— → 118K
Qwen: Qwen3 Coder 30B A3B Instruct最大输出262K → 236K
Mistral: Codestral 2508最大输出— → 205K
OpenAI: gpt-oss-20b最大输出131K → 118K
OpenAI: gpt-oss-120b最大输出131K → 118K
Mistral: Mistral Medium 3.1最大输出— → 105K
DeepSeek: DeepSeek V3.1最大输出161K → 145K
Nous: Hermes 4 405B最大输出— → 118K
Nous: Hermes 4 70B最大输出— → 118K
Qwen: Qwen3 Next 80B A3B Instruct最大输出262K → 236K
TheDrummer: Cydonia 24B V4.1最大输出131K → 118K
IBM: Granite 4.0 Micro最大输出131K → 118K
Mistral: Voxtral Small 24B 2507最大输出— → 26K
AllenAI: Olmo 3 32B Think最大输出66K → 59K
DeepSeek: DeepSeek V3.2最大输出164K → 147K
Mistral: Mistral Large 3 2512最大输出— → 210K
Mistral: Ministral 3 3B 2512最大输出— → 105K
Mistral: Ministral 3 8B 2512最大输出— → 210K
Mistral: Ministral 3 14B 2512最大输出— → 210K
Mistral: Devstral 2 2512最大输出— → 210K
NVIDIA: Nemotron 3 Nano 30B A3B最大输出262K → 236K
Upstage: Solar Pro 3最大输出131K → 118K
MoonshotAI: Kimi K2.5最大输出— → 236K
Qwen: Qwen3 Coder Next最大输出262K → 236K
Qwen: Qwen3.5 397B A17B最大输出262K → 236K
Qwen: Qwen3.5-122B-A10B最大输出262K → 236K
Qwen: Qwen3.5-35B-A3B最大输出262K → 236K
Google: Nano Banana 2 (Gemini 3.1 Flash Image Preview)最大输出66K → 59K
Qwen: Qwen3.5-9B最大输出262K → 236K
Mistral: Mistral Small 4最大输出— → 210K
MiniMax: MiniMax M2.7最大输出— → 131K
Reka Edge最大输出16K → 15K
SpaceXAI: Grok 4.20最大输出— → 1.8M
SpaceXAI: Grok 4.20 Multi-Agent最大输出— → 1.8M
Arcee AI: Trinity Large Thinking最大输出262K → 236K
Google: Gemma 4 31B最大输出262K → 236K
Z.ai: GLM 5.1最大输出128K → 182K
Z.ai: GLM 5.1缓存读$0.18 → $0.23+30%
Z.ai: GLM 5.1输出价$3.04 → $3.96+30%
Z.ai: GLM 5.1输入价$0.97 → $1.26+30%
MoonshotAI: Kimi K2.6最大输出— → 236K
Tencent: Hy3 preview最大输出— → 236K
DeepSeek: DeepSeek V4 Flash 0423缓存读$0.01 → $0.02+38%
DeepSeek: DeepSeek V4 Flash 0423输出价$0.11 → $0.15+37%
DeepSeek: DeepSeek V4 Flash 0423输入价$0.06 → $0.08+37%
DeepSeek: DeepSeek V4 Pro 0423缓存读$0.04 → $0.05+6%
DeepSeek: DeepSeek V4 Pro 0423输出价$1.04 → $1.11+7%
DeepSeek: DeepSeek V4 Pro 0423输入价$0.52 → $0.56+6%
Qwen: Qwen3.6 27B缓存读$0.12 → —
Qwen: Qwen3.6 35B A3B最大输出262K → 236K
Mistral: Mistral Medium 3.5最大输出— → 210K
IBM: Granite 4.1 8B最大输出131K → 118K
SpaceXAI: Grok 4.3最大输出— → 900K
SpaceXAI: Grok Build 0.1最大输出— → 230K
StepFun: Step 3.7 Flash最大输出256K → 230K
NVIDIA: Nemotron 3 Ultra最大输出— → 461K
Nex AGI: Nex-N2-Pro最大输出262K → 236K
MoonshotAI: Kimi K2.7 Code最大输出262K → 236K
Z.ai: GLM 5.2最大输出131K → 262K
Z.ai: GLM 5.2缓存读$0.19 → $0.22+14%
Z.ai: GLM 5.2输出价$3.04 → $3.74+23%
Z.ai: GLM 5.2输入价$0.97 → $1.19+23%
Nex AGI: Nex-N2-Mini最大输出262K → 236K
Google: Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)最大输出66K → 59K
SpaceXAI: Grok 4.5最大输出— → 450K
Meta: Muse Spark 1.1最大输出— → 944K
MoonshotAI: Kimi K3最大输出1.05M → 944K
Thinking Machines: Inkling最大输出— → 262K
Thinking Machines: Inkling缓存读$0.17 → $0.16-6%
Thinking Machines: Inkling输入价$1.00 → $0.95-5%
DeepSeek: DeepSeek V4 Flash 0731最大输出131K → 944K
DeepSeek: DeepSeek V4 Flash 0731缓存读$0.01 → $0.01-39%
DeepSeek: DeepSeek V4 Flash 0731输出价$0.13 → $0.08-39%
DeepSeek: DeepSeek V4 Flash 0731输入价$0.07 → $0.04-39%
Meta: Muse Spark 1.2最大输出— → 944K
Meta: Muse Glimmer 30B最大输出— → 118K
SpaceXAI: Grok 4.6最大输出— → 450K
DeepSeek: DeepSeek V4 Pro 0813最大输出— → 944K
ByteDance Seed: Seed 2.1 Turbo最大输出262K → 236K
Qwen: Qwen3.8 27B缓存写— → $0.53
Qwen: Qwen3.8 27B缓存读$0.05 → $0.09+70%
Qwen: Qwen3.8 27B输出价$3.00 → $2.55-15%
Qwen: Qwen3.8 27B输入价$0.40 → $0.42+6%
Meta: Muse Spark 1.2 Contributor最大输出— → 944K

2026-08-24

ReMM SLERP 13B最大输出6K → 4K
TheDrummer: UnslopNemo 12B最大输出1.02M → 33K
Qwen: Qwen3 30B A3B最大输出8K → 16K
Qwen: Qwen3 30B A3B输出价$0.52 → $0.50-4%
Qwen: Qwen3 30B A3B输入价$0.13 → $0.12-8%
Qwen: Qwen3 235B A22B Thinking 2507上下文窗口262K → 131K
Mistral: Devstral 2 2512缓存读$0.04 → $0.04+10%
Mistral: Devstral 2 2512输出价$2.00 → $2.20+10%
Mistral: Devstral 2 2512输入价$0.40 → $0.44+10%
MoonshotAI: Kimi K2.5最大输出262K → —
MoonshotAI: Kimi K2.5缓存读$0.07 → $0.10+43%
MoonshotAI: Kimi K2.5输出价$2.25 → $3.00+33%
MoonshotAI: Kimi K2.5输入价$0.45 → $0.60+33%
Qwen: Qwen3.5 397B A17B最大输出66K → 262K
Qwen: Qwen3.5 397B A17B缓存读— → $0.30
Qwen: Qwen3.5 397B A17B输出价$2.34 → $3.60+54%
Qwen: Qwen3.5 397B A17B输入价$0.39 → $0.50+28%
Qwen: Qwen3.5-122B-A10B最大输出66K → 262K
MiniMax: MiniMax M2.7缓存读$0.05 → $0.06+25%
MiniMax: MiniMax M2.7输出价$0.96 → $1.20+25%
MiniMax: MiniMax M2.7输入价$0.24 → $0.30+25%
MoonshotAI: Kimi K2.6最大输出262K → —
MoonshotAI: Kimi K2.6缓存读$0.09 → $0.16+75%
MoonshotAI: Kimi K2.6输出价$2.28 → $4.00+75%
MoonshotAI: Kimi K2.6输入价$0.54 → $0.95+75%
DeepSeek: DeepSeek V4 Flash 0423缓存读$0.01 → $0.01+14%
DeepSeek: DeepSeek V4 Flash 0423输出价$0.10 → $0.11+15%
DeepSeek: DeepSeek V4 Flash 0423输入价$0.05 → $0.06+15%
DeepSeek: DeepSeek V4 Pro 0423缓存读$0.03 → $0.04+31%
DeepSeek: DeepSeek V4 Pro 0423输出价$0.79 → $1.04+32%
DeepSeek: DeepSeek V4 Pro 0423输入价$0.40 → $0.52+32%
Qwen: Qwen3.6 27B最大输出262K → 82K
Qwen: Qwen3.6 27B输出价$3.60 → $3.20-11%
Qwen: Qwen3.6 27B输入价$0.60 → $0.32-47%
MoonshotAI: Kimi K2.7 Code缓存读$0.17 → $0.19+12%
Thinking Machines: Inkling最大输出262K → —
Thinking Machines: Inkling缓存读$0.16 → $0.17+6%
Thinking Machines: Inkling输入价$0.95 → $1.00+5%
DeepSeek: DeepSeek V4 Flash 0731最大输出384K → 131K
DeepSeek: DeepSeek V4 Flash 0731缓存读$0.02 → $0.01-17%
DeepSeek: DeepSeek V4 Flash 0731输出价$0.18 → $0.13-27%
DeepSeek: DeepSeek V4 Flash 0731输入价$0.08 → $0.07-18%

2026-08-23

OpenAI: gpt-oss-120b缓存读$0.03 → —
MiniMax: MiniMax M2.7最大输出131K → —
MiniMax: MiniMax M2.7缓存读$0.06 → $0.05-20%
MiniMax: MiniMax M2.7输出价$1.20 → $0.96-20%
MiniMax: MiniMax M2.7输入价$0.30 → $0.24-20%
DeepSeek: DeepSeek V4 Flash 0423缓存读$0.01 → $0.01-15%
DeepSeek: DeepSeek V4 Flash 0423输出价$0.11 → $0.10-15%
DeepSeek: DeepSeek V4 Flash 0423输入价$0.06 → $0.05-15%
DeepSeek: DeepSeek V4 Pro 0423缓存读$0.03 → $0.03-4%
DeepSeek: DeepSeek V4 Pro 0423输出价$0.83 → $0.79-4%
DeepSeek: DeepSeek V4 Pro 0423输入价$0.41 → $0.40-4%
MoonshotAI: Kimi K3最大输出— → 1.05M

2026-08-22

Mistral: Mistral Small 3.2 24B上下文窗口256K → 131K
Mistral: Mistral Small 3.2 24B输出价$0.25 → $0.20-20%
Mistral: Mistral Small 3.2 24B输入价$0.09 → $0.07-20%
OpenAI: gpt-oss-120b输入价$0.03 → $0.04+23%
DeepSeek: DeepSeek V3.1最大输出33K → 161K
DeepSeek: DeepSeek V3.1缓存读$0.13 → $0.55+323%
DeepSeek: DeepSeek V3.1输出价$0.95 → $1.65+74%
DeepSeek: DeepSeek V3.1输入价$0.25 → $0.55+120%
Qwen: Qwen3 Next 80B A3B Instruct最大输出16K → 262K
Qwen: Qwen3 Next 80B A3B Instruct缓存读— → $0.07
Qwen: Qwen3 Next 80B A3B Instruct输入价$0.09 → $0.10+11%
DeepSeek: DeepSeek V3.2最大输出66K → 164K
DeepSeek: DeepSeek V3.2缓存读$0.13 → $0.13-3%
DeepSeek: DeepSeek V3.2输出价$0.40 → $0.38-5%
DeepSeek: DeepSeek V3.2输入价$0.27 → $0.26-3%
MiniMax: MiniMax M2.5最大输出33K → 128K
MiniMax: MiniMax M2.5缓存读$0.03 → $0.03-10%
MiniMax: MiniMax M2.5输出价$0.95 → $1.08+14%
Qwen: Qwen3.5-122B-A10B最大输出262K → 66K
MoonshotAI: Kimi K2.6缓存读$0.10 → $0.09-7%
MoonshotAI: Kimi K2.6输出价$2.44 → $2.28-7%
MoonshotAI: Kimi K2.6输入价$0.58 → $0.54-7%
DeepSeek: DeepSeek V4 Flash 0423缓存读$0.02 → $0.01-28%
DeepSeek: DeepSeek V4 Flash 0423输出价$0.16 → $0.11-28%
DeepSeek: DeepSeek V4 Flash 0423输入价$0.08 → $0.06-28%
DeepSeek: DeepSeek V4 Pro 0423最大输出393K → 384K
DeepSeek: DeepSeek V4 Pro 0423缓存读$0.14 → $0.03-74%
DeepSeek: DeepSeek V4 Pro 0423输出价$3.20 → $0.83-74%
DeepSeek: DeepSeek V4 Pro 0423输入价$1.60 → $0.41-74%
OpenAI: GPT-5.6 Sol缓存写$3.13 → $2.50-20%
OpenAI: GPT-5.6 Sol缓存读$0.25 → $0.20-20%
OpenAI: GPT-5.6 Sol输出价$15.00 → $10.00-33%
OpenAI: GPT-5.6 Sol输入价$2.50 → $2.00-20%
OpenAI: GPT-5.6 Sol Pro缓存写$3.13 → $2.50-20%
OpenAI: GPT-5.6 Sol Pro缓存读$0.25 → $0.20-20%
OpenAI: GPT-5.6 Sol Pro输出价$15.00 → $10.00-33%
OpenAI: GPT-5.6 Sol Pro输入价$2.50 → $2.00-20%
Meta: Muse Glimmer 30B最大输出131K → —
Meta: Muse Glimmer 30B输出价$1.10 → $1.50+36%
Meta: Muse Glimmer 30B输入价$0.30 → $0.35+17%
DeepSeek: DeepSeek V4 Pro 0813缓存读$0.04 → $0.04-6%
DeepSeek: DeepSeek V4 Pro 0813输出价$3.56 → $3.37-6%
DeepSeek: DeepSeek V4 Pro 0813输入价$1.19 → $1.12-6%
Qwen: Qwen3.8 27B输出价$3.20 → $3.00-6%
Qwen: Qwen3.8 27B输入价$0.45 → $0.40-11%

2026-08-21

MiniMax: MiniMax M2.5最大输出— → 33K
MiniMax: MiniMax M2.5缓存读$0.06 → $0.03-50%
MiniMax: MiniMax M2.5输出价$0.90 → $0.95+6%
MiniMax: MiniMax M2.5输入价$0.23 → $0.27+20%
Google: Gemma 4 31B最大输出16K → 262K
Google: Gemma 4 31B缓存读$0.05 → $0.10+100%
Google: Gemma 4 31B输入价$0.09 → $0.10+11%
MoonshotAI: Kimi K2.6缓存读$0.09 → $0.10+7%
MoonshotAI: Kimi K2.6输出价$2.28 → $2.44+7%
MoonshotAI: Kimi K2.6输入价$0.54 → $0.58+7%
DeepSeek: DeepSeek V4 Flash 0423缓存读$0.02 → $0.02+5%
DeepSeek: DeepSeek V4 Flash 0423输出价$0.15 → $0.16+5%
DeepSeek: DeepSeek V4 Flash 0423输入价$0.08 → $0.08+5%
DeepSeek: DeepSeek V4 Pro 0423缓存读$0.02 → $0.14+514%
MoonshotAI: Kimi K2.7 Code缓存读$0.15 → $0.17+13%
MoonshotAI: Kimi K2.7 Code输出价$3.50 → $3.40-3%
MoonshotAI: Kimi K2.7 Code输入价$0.71 → $0.67-6%
DeepSeek: DeepSeek V4 Flash 0731最大输出393K → 384K
DeepSeek: DeepSeek V4 Flash 0731缓存读$0.03 → $0.02-43%
DeepSeek: DeepSeek V4 Flash 0731输出价$0.28 → $0.18-36%
DeepSeek: DeepSeek V4 Flash 0731输入价$0.14 → $0.08-43%
Meta: Muse Glimmer 30B最大输出— → 131K
Meta: Muse Glimmer 30B输出价$1.50 → $1.10-27%
Meta: Muse Glimmer 30B输入价$0.35 → $0.30-14%
Qwen: Qwen3.8 2.4T A95B最大输出262K → 131K

2026-08-20

DeepSeek: DeepSeek V3 0324缓存读$0.14 → —
Qwen: Qwen3 Next 80B A3B Instruct缓存读$0.07 → —
DeepSeek: DeepSeek V3.1 Terminus最大输出164K → 33K
DeepSeek: DeepSeek V3.1 Terminus缓存读— → $0.14
Google: Gemma 4 31B最大输出262K → 16K
Google: Gemma 4 31B缓存读$0.10 → $0.05-50%
Google: Gemma 4 31B输入价$0.10 → $0.09-10%
DeepSeek: DeepSeek V4 Flash 0423缓存读$0.01 → $0.02+2%
DeepSeek: DeepSeek V4 Flash 0423输出价$0.15 → $0.15+2%
DeepSeek: DeepSeek V4 Flash 0423输入价$0.07 → $0.08+2%
DeepSeek: DeepSeek V4 Pro 0423输出价$2.88 → $3.20+11%
DeepSeek: DeepSeek V4 Pro 0423输入价$1.44 → $1.60+11%
Qwen: Qwen3.6 27B最大输出66K → 262K
Qwen: Qwen3.6 27B缓存读$0.03 → $0.12+300%
Qwen: Qwen3.6 27B输出价$2.00 → $3.60+80%
Qwen: Qwen3.6 27B输入价$0.30 → $0.60+100%
Z.ai: GLM 5.2缓存读$0.01 → $0.19+1386%
Z.ai: GLM 5.2输出价$0.22 → $3.04+1280%
Z.ai: GLM 5.2输入价$0.07 → $0.97+1280%
Thinking Machines: Inkling Small上下文窗口524K → 1.05M
NVIDIA: Nemotron 3.5 Lightning上下文窗口1M → 262K
DeepSeek: DeepSeek V4 Pro 0813最大输出384K → —
DeepSeek: DeepSeek V4 Pro 0813缓存读$0.02 → $0.04+80%
DeepSeek: DeepSeek V4 Pro 0813输出价$1.98 → $3.56+80%
DeepSeek: DeepSeek V4 Pro 0813输入价$0.66 → $1.19+80%

2026-08-19

DeepSeek: DeepSeek V3 0324最大输出66K → 164K
DeepSeek: DeepSeek V3 0324输出价$1.12 → $1.00-11%
DeepSeek: DeepSeek V3 0324输入价$0.27 → $0.25-7%
Qwen: Qwen3 Next 80B A3B Instruct最大输出262K → 16K
Qwen: Qwen3 Next 80B A3B Instruct输入价$0.10 → $0.09-10%
MoonshotAI: Kimi K2.5缓存读$0.10 → $0.07-26%
MoonshotAI: Kimi K2.5输出价$2.85 → $2.25-21%
MoonshotAI: Kimi K2.5输入价$0.57 → $0.45-21%
MiniMax: MiniMax M2.5最大输出197K → —
MiniMax: MiniMax M2.5缓存读$0.05 → $0.06+20%
MiniMax: MiniMax M2.5输入价$0.22 → $0.23+2%
Qwen: Qwen3.5-122B-A10B最大输出82K → 262K
Qwen: Qwen3.5-122B-A10B输出价$2.40 → $2.08-13%
Qwen: Qwen3.5-122B-A10B输入价$0.29 → $0.26-10%
Qwen: Qwen3.5-35B-A3B最大输出66K → 262K
Qwen: Qwen3.5-35B-A3B缓存读$0.23 → $0.25+11%
Qwen: Qwen3.5-35B-A3B输出价$1.80 → $1.25-31%
Qwen: Qwen3.5-35B-A3B输入价$0.23 → $0.25+11%
MoonshotAI: Kimi K2.6缓存读$0.09 → $0.09-3%
MoonshotAI: Kimi K2.6输出价$2.36 → $2.28-3%
MoonshotAI: Kimi K2.6输入价$0.56 → $0.54-3%
DeepSeek: DeepSeek V4 Pro 0423最大输出384K → 393K
DeepSeek: DeepSeek V4 Pro 0423输出价$1.98 → $2.88+45%
DeepSeek: DeepSeek V4 Pro 0423输入价$0.66 → $1.44+118%
Qwen: Qwen3.6 27B最大输出131K → 66K
Qwen: Qwen3.6 27B缓存读— → $0.03
Qwen: Qwen3.6 27B输出价$2.40 → $2.00-17%
Qwen: Qwen3.6 27B输入价$0.29 → $0.30+4%
Z.ai: GLM 5.2最大输出262K → 131K
Google: Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)最大输出66K → 66K
Qwen: Qwen3.8 27B上下文窗口262K → 1M

2026-08-18

DeepSeek: DeepSeek V3.1 Terminus缓存读$0.13 → —
DeepSeek: DeepSeek V4 Flash 0423缓存读$0.01 → $0.01+3%
DeepSeek: DeepSeek V4 Flash 0423输出价$0.14 → $0.15+4%
DeepSeek: DeepSeek V4 Flash 0423输入价$0.07 → $0.07+4%
Qwen: Qwen3.6 27B缓存读$0.03 → —
Z.ai: GLM 5.2最大输出131K → 262K

2026-08-17

DeepSeek: DeepSeek V3.1 Terminus最大输出33K → 164K
DeepSeek: DeepSeek V3.1 Terminus输出价$0.95 → $1.00+5%
Qwen: Qwen3 VL 235B A22B Instruct缓存读— → $0.10
Qwen: Qwen3 VL 235B A22B Instruct输出价$1.04 → $1.90+83%
Qwen: Qwen3 VL 235B A22B Instruct输入价$0.26 → $0.21-19%
Z.ai: GLM 4.6缓存读$0.11 → $0.10-9%
Z.ai: GLM 4.6输出价$2.20 → $2.00-9%
Z.ai: GLM 4.6输入价$0.55 → $0.50-9%
NVIDIA: Nemotron 3 Nano 30B A3B最大输出228K → 262K
NVIDIA: Nemotron 3 Nano 30B A3B缓存读$0.03 → $0.03+20%
Google: Gemma 4 26B A4B 缓存读$0.05 → —
MoonshotAI: Kimi K2.6缓存读$0.09 → $0.09+4%
MoonshotAI: Kimi K2.6输出价$2.28 → $2.36+4%
MoonshotAI: Kimi K2.6输入价$0.54 → $0.56+4%
DeepSeek: DeepSeek V4 Flash 0423缓存读$0.01 → $0.01+19%
DeepSeek: DeepSeek V4 Flash 0423输出价$0.12 → $0.14+19%
DeepSeek: DeepSeek V4 Flash 0423输入价$0.06 → $0.07+19%
DeepSeek: DeepSeek V4 Pro 0423最大输出393K → 384K
DeepSeek: DeepSeek V4 Pro 0423缓存读$0.10 → $0.02-78%
DeepSeek: DeepSeek V4 Pro 0423输出价$2.34 → $1.98-15%
DeepSeek: DeepSeek V4 Pro 0423输入价$1.17 → $0.66-43%
Qwen: Qwen3.6 27B最大输出66K → 131K
Qwen: Qwen3.6 27B输出价$2.00 → $2.40+20%
Qwen: Qwen3.6 27B输入价$0.30 → $0.29-4%
Z.ai: GLM 5.2最大输出128K → 131K
OpenAI: GPT-5.6 Sol缓存写$6.25 → $3.13-50%
OpenAI: GPT-5.6 Sol缓存读$0.50 → $0.25-50%
OpenAI: GPT-5.6 Sol输出价$30.00 → $15.00-50%
OpenAI: GPT-5.6 Sol输入价$5.00 → $2.50-50%
OpenAI: GPT-5.6 Sol Pro缓存写$6.25 → $3.13-50%
OpenAI: GPT-5.6 Sol Pro缓存读$0.50 → $0.25-50%
OpenAI: GPT-5.6 Sol Pro输出价$30.00 → $15.00-50%
OpenAI: GPT-5.6 Sol Pro输入价$5.00 → $2.50-50%
OpenAI: GPT-5.6 Terra缓存写$1.25 → $2.50+100%
OpenAI: GPT-5.6 Terra缓存读$0.10 → $0.20+100%
OpenAI: GPT-5.6 Terra输出价$6.00 → $12.00+100%
OpenAI: GPT-5.6 Terra输入价$1.00 → $2.00+100%
OpenAI: GPT-5.6 Terra Pro缓存写$1.25 → $2.50+100%
OpenAI: GPT-5.6 Terra Pro缓存读$0.10 → $0.20+100%
OpenAI: GPT-5.6 Terra Pro输出价$6.00 → $12.00+100%
OpenAI: GPT-5.6 Terra Pro输入价$1.00 → $2.00+100%
OpenAI: GPT-5.6 Luna缓存写$0.13 → $0.25+100%
OpenAI: GPT-5.6 Luna缓存读$0.01 → $0.02+100%
OpenAI: GPT-5.6 Luna输出价$0.60 → $1.20+100%
OpenAI: GPT-5.6 Luna输入价$0.10 → $0.20+100%
OpenAI: GPT-5.6 Luna Pro缓存写$0.13 → $0.25+100%
OpenAI: GPT-5.6 Luna Pro缓存读$0.01 → $0.02+100%
OpenAI: GPT-5.6 Luna Pro输出价$0.60 → $1.20+100%
OpenAI: GPT-5.6 Luna Pro输入价$0.10 → $0.20+100%
DeepSeek: DeepSeek V4 Flash 0731上下文窗口1.05M → 1.31M
DeepSeek: DeepSeek V4 Pro 0813缓存读$0.00 → $0.02+511%

2026-08-16

Qwen: Qwen2.5 VL 72B Instruct最大输出— → 128K
Qwen: Qwen2.5 VL 72B Instruct缓存读— → $0.40
Qwen: Qwen2.5 VL 72B Instruct输出价$0.75 → $1.00+33%
Qwen: Qwen2.5 VL 72B Instruct输入价$0.25 → $0.80+220%
Qwen: Qwen3 30B A3B最大输出16K → 8K
Qwen: Qwen3 30B A3B输出价$0.50 → $0.52+4%
Qwen: Qwen3 30B A3B输入价$0.12 → $0.13+8%
Google: Gemma 4 26B A4B 最大输出262K → 16K
Google: Gemma 4 26B A4B 输出价$0.40 → $0.34-15%
Google: Gemma 4 26B A4B 输入价$0.12 → $0.07-42%
DeepSeek: DeepSeek V4 Flash 0423缓存读$0.01 → $0.01-7%
DeepSeek: DeepSeek V4 Flash 0423输出价$0.13 → $0.12-7%
DeepSeek: DeepSeek V4 Flash 0423输入价$0.06 → $0.06-7%
Qwen: Qwen3.6 27B最大输出262K → 66K
Qwen: Qwen3.6 27B缓存读$0.12 → $0.03-75%
Qwen: Qwen3.6 27B输出价$3.60 → $2.00-44%
Qwen: Qwen3.6 27B输入价$0.60 → $0.30-50%
Qwen: Qwen3.6 35B A3B输入价$0.15 → $0.14-7%
Z.ai: GLM 5.2最大输出131K → 128K
NVIDIA: Nemotron 3.5 Lightning最大输出262K → 131K
NVIDIA: Nemotron 3.5 Lightning缓存读$0.05 → $0.04-20%
NVIDIA: Nemotron 3.5 Lightning输出价$0.25 → $0.20-20%
NVIDIA: Nemotron 3.5 Lightning输入价$0.10 → $0.08-20%
DeepSeek: DeepSeek V4 Pro 0813输出价$0.87 → $1.98+128%
DeepSeek: DeepSeek V4 Pro 0813输入价$0.43 → $0.66+52%
Qwen: Qwen3.8 27B缓存读— → $0.05

2026-08-15

Qwen: Qwen3.5 397B A17B缓存读$0.30 → —
Z.ai: GLM 5.1最大输出131K → 128K
Z.ai: GLM 5.1缓存读$0.26 → $0.18-31%
Z.ai: GLM 5.1输出价$4.40 → $3.04-31%
Z.ai: GLM 5.1输入价$1.40 → $0.97-31%
MoonshotAI: Kimi K2.6缓存读$0.09 → $0.09-3%
MoonshotAI: Kimi K2.6输出价$2.36 → $2.28-3%
MoonshotAI: Kimi K2.6输入价$0.56 → $0.54-3%
Tencent: Hy3 preview缓存读$0.02 → $0.06+186%
Tencent: Hy3 preview输出价$0.21 → $0.60+186%
Tencent: Hy3 preview输入价$0.06 → $0.18+186%
DeepSeek: DeepSeek V4 Flash 0423最大输出393K → 384K
DeepSeek: DeepSeek V4 Flash 0423缓存读$0.03 → $0.01-54%
DeepSeek: DeepSeek V4 Flash 0423输出价$0.28 → $0.13-54%
DeepSeek: DeepSeek V4 Flash 0423输入价$0.14 → $0.06-54%
Z.ai: GLM 5.2最大输出262K → 131K

2026-08-14

Meta: Llama 4 Maverick最大输出— → 16K
Meta: Llama 4 Maverick输出价$0.70 → $0.80+15%
Z.ai: GLM 4.6缓存读$0.10 → $0.11+10%
Z.ai: GLM 4.6输出价$2.00 → $2.20+10%
Z.ai: GLM 4.6输入价$0.50 → $0.55+10%
Qwen: Qwen3 VL 30B A3B Instruct最大输出16K → 33K
Qwen: Qwen3 VL 30B A3B Instruct输出价$0.60 → $0.52-13%
Qwen: Qwen3 VL 30B A3B Instruct输入价$0.15 → $0.13-13%
Z.ai: GLM 5最大输出131K → 128K
Z.ai: GLM 5缓存读$0.20 → $0.12-40%
Z.ai: GLM 5输出价$2.55 → $1.92-25%
Z.ai: GLM 5输入价$0.95 → $0.60-37%
Qwen: Qwen3.5 397B A17B最大输出262K → 66K
Qwen: Qwen3.5 397B A17B输出价$3.60 → $2.34-35%
Qwen: Qwen3.5 397B A17B输入价$0.50 → $0.39-22%
Qwen: Qwen3.5-35B-A3B最大输出262K → 66K
Qwen: Qwen3.5-35B-A3B缓存读$0.25 → $0.23-10%
Qwen: Qwen3.5-35B-A3B输出价$1.25 → $1.80+44%
Qwen: Qwen3.5-35B-A3B输入价$0.25 → $0.23-10%
Z.ai: GLM 5.1缓存读$0.18 → $0.26+47%
Z.ai: GLM 5.1输出价$2.99 → $4.40+47%
Z.ai: GLM 5.1输入价$0.95 → $1.40+47%
MoonshotAI: Kimi K2.7 Code输出价$3.40 → $3.50+3%
MoonshotAI: Kimi K2.7 Code输入价$0.67 → $0.71+6%
Z.ai: GLM 5.2最大输出131K → 262K
Google: Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)最大输出66K → 66K
DeepSeek: DeepSeek V4 Flash 0731最大输出384K → 393K
DeepSeek: DeepSeek V4 Flash 0731缓存读$0.02 → $0.03+75%
DeepSeek: DeepSeek V4 Flash 0731输出价$0.18 → $0.28+56%
DeepSeek: DeepSeek V4 Flash 0731输入价$0.08 → $0.14+75%
Qwen: Qwen3.8 2.4T A95B上下文窗口1M → 1.05M

2026-08-13

Magnum v4 72B最大输出2K → 4K
Magnum v4 72B上下文窗口16K → 33K
DeepSeek: R1上下文窗口164K → 64K
Qwen: Qwen3 Next 80B A3B Instruct最大输出16K → 262K
Qwen: Qwen3 Next 80B A3B Instruct缓存读— → $0.07
Qwen: Qwen3 Next 80B A3B Instruct输入价$0.09 → $0.10+11%
Qwen: Qwen3 Next 80B A3B Thinking最大输出262K → 33K

价格数据参考 OpenRouter 及各厂商公开页面整理,以美元 / 百万 tokens 计,可能滞后或有误,请以官方页面为准。不构成任何采购或投资建议。