价格变动时间线
各家大模型 API 的调价 / 降价记录,按日期倒序 · 数据更新于 2026-10-23
📊 全部数据 · 总览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
2026-06-27
- Qwen: Qwen3.6 27B输出价$2.65 → $2.38-10%
- Qwen: Qwen3.6 27B输入价$0.29 → $0.26-10%
2026-06-26
- DeepSeek: DeepSeek V3.2缓存读— → $0.02
- Qwen: Qwen3.6 27B输出价$3.17 → $2.65-16%
2026-06-25
- MiniMax: MiniMax M2.5缓存读$0.05 → —
- OpenAI: gpt-oss-120b最大输出— → 131K
- OpenAI: gpt-oss-120b输出价$0.18 → $0.15-17%
- OpenAI: gpt-oss-120b输入价$0.04 → $0.03-23%
- MiniMax: MiniMax M2.5输出价$0.90 → $0.48-47%
- MiniMax: MiniMax M2.5输入价$0.15 → $0.12-20%
- NVIDIA: Nemotron 3 Super最大输出— → 16K
- NVIDIA: Nemotron 3 Super输出价$0.45 → $0.40-11%
- NVIDIA: Nemotron 3 Super输入价$0.09 → $0.09-6%
- MiniMax: MiniMax M2.7输出价$0.96 → $0.72-25%
- MiniMax: MiniMax M2.7输入价$0.24 → $0.18-25%
- DeepSeek: DeepSeek V4 Flash最大输出— → 66K
- DeepSeek: DeepSeek V4 Flash缓存读$0.03 → $0.02-25%
- DeepSeek: DeepSeek V4 Flash输出价$0.22 → $0.18-20%
- DeepSeek: DeepSeek V4 Flash输入价$0.09 → $0.09+1%
2026-06-24
- Google: Nano Banana (Gemini 2.5 Flash Image)最大输出8K → 33K
- MoonshotAI: Kimi K2 Thinking缓存读— → $0.60
- Qwen: Qwen3.5-35B-A3B最大输出262K → 82K
- Qwen: Qwen3.5-35B-A3B缓存读— → $0.05
- Z.ai: GLM 5.1最大输出66K → —
- Z.ai: GLM 5.1缓存读$0.49 → $0.18-63%
- DeepSeek: DeepSeek V4 Flash最大输出66K → —
- DeepSeek: DeepSeek V4 Flash缓存读$0.02 → $0.03+33%
- DeepSeek: DeepSeek V4 Flash输出价$0.18 → $0.22+24%
- DeepSeek: DeepSeek V4 Flash输入价$0.09 → $0.09-1%
- Google: Nano Banana 2 (Gemini 3.1 Flash Image)最大输出66K → 33K
- Xiaomi: MiMo-V2.5缓存读$0.00 → —
2026-06-23
- MiniMax: MiniMax M2.7缓存读$0.05 → —
- Xiaomi: MiMo-V2.5最大输出131K → —
- Xiaomi: MiMo-V2.5输入价$0.14 → $0.10-25%
- Nex AGI: Nex-N2-Pro缓存读$0.25 → $0.03-90%
- TheDrummer: Rocinante 12B最大输出33K → —
- TheDrummer: Rocinante 12B输出价$0.43 → $0.50+16%
- TheDrummer: Rocinante 12B输入价$0.17 → $0.25+47%
- Microsoft: Phi 4输入价$0.07 → $0.07+8%
- MiniMax: MiniMax M2.7最大输出131K → 197K
- MiniMax: MiniMax M2.7输出价$1.00 → $0.96-4%
- MiniMax: MiniMax M2.7输入价$0.25 → $0.24-4%
- Nex AGI: Nex-N2-Pro输出价$2.50 → $1.00-60%
- Nex AGI: Nex-N2-Pro输入价$0.50 → $0.25-50%
- MoonshotAI: Kimi K2.7 Code最大输出262K → 16K
- MoonshotAI: Kimi K2.7 Code缓存读$0.13 → $0.15+16%
- MoonshotAI: Kimi K2.7 Code输出价$3.07 → $3.50+14%
- MoonshotAI: Kimi K2.7 Code输入价$0.61 → $0.74+21%
- Z.ai: GLM 5.2最大输出— → 33K
- Z.ai: GLM 5.2缓存读$0.18 → $0.18-1%
- Z.ai: GLM 5.2输出价$3.08 → $3.00-3%
- Z.ai: GLM 5.2输入价$0.98 → $0.95-3%
2026-06-22
- OpenAI: GPT-4o缓存读$1.25 → —
- MoonshotAI: Kimi K2 Thinking缓存读$0.60 → —
- Qwen: Qwen3.5-35B-A3B缓存读$0.05 → —
- MoonshotAI: Kimi K2.6最大输出66K → 262K
- MoonshotAI: Kimi K2.6缓存读$0.33 → $0.14-56%
- MoonshotAI: Kimi K2.6输出价$3.50 → $3.41-3%
- Z.ai: GLM 5.2最大输出33K → —
- Z.ai: GLM 5.2缓存读$0.18 → $0.18+1%
- Z.ai: GLM 5.2输出价$4.00 → $3.08-23%
- Z.ai: GLM 5.2输入价$1.00 → $0.98-2%
2026-06-21
- OpenAI: GPT-4.1 Mini最大输出— → 33K
- OpenAI: GPT-5 Mini最大输出— → 128K
- OpenAI: GPT-5 Mini缓存读$0.03 → $0.03-17%
- OpenAI: GPT-5.1 Chat最大输出16K → 32K
- OpenAI: GPT-5.1 Chat缓存读$0.13 → $0.13+4%
- OpenAI: GPT-5.2 Chat最大输出32K → 16K
- Qwen: Qwen3.5-35B-A3B最大输出82K → 262K
- Google: Nano Banana 2 (Gemini 3.1 Flash Image Preview)最大输出66K → 33K
- Qwen: Qwen3.5-9B最大输出33K → 262K
- Qwen: Qwen3.5-9B上下文窗口256K → 262K
- Kwaipilot: KAT-Coder-Pro V2最大输出144K → 80K
- Kwaipilot: KAT-Coder-Pro V2上下文窗口262K → 256K
- Z.ai: GLM 5.1最大输出— → 66K
- Z.ai: GLM 5.1缓存读$0.18 → $0.49+169%
- MoonshotAI: Kimi K2.6最大输出262K → 66K
- MiniMax: MiniMax M3最大输出— → 512K
- Google: Nano Banana Pro (Gemini 3 Pro Image)上下文窗口131K → 66K
- Google: Nano Banana 2 (Gemini 3.1 Flash Image)最大输出33K → 66K
- OpenAI: GPT-4o缓存读— → $1.25
- OpenAI: GPT-4.1 Mini最大输出33K → —
- OpenAI: GPT-5 Mini最大输出128K → —
- OpenAI: GPT-5 Mini缓存读$0.03 → $0.03+20%
- MoonshotAI: Kimi K2 Thinking缓存读— → $0.60
- OpenAI: GPT-5.1 Chat最大输出32K → 16K
- OpenAI: GPT-5.1 Chat缓存读$0.13 → $0.13-4%
- OpenAI: GPT-5.2 Chat最大输出16K → 32K
- Qwen: Qwen3.5-35B-A3B最大输出262K → 82K
- Qwen: Qwen3.5-35B-A3B缓存读— → $0.05
- Google: Nano Banana 2 (Gemini 3.1 Flash Image Preview)最大输出33K → 66K
- Qwen: Qwen3.5-9B最大输出262K → 33K
- Qwen: Qwen3.5-9B上下文窗口262K → 256K
- Kwaipilot: KAT-Coder-Pro V2最大输出80K → 144K
- Kwaipilot: KAT-Coder-Pro V2上下文窗口256K → 262K
- Z.ai: GLM 5.1最大输出66K → —
- Z.ai: GLM 5.1缓存读$0.49 → $0.18-63%
- MiniMax: MiniMax M3最大输出512K → —
- Z.ai: GLM 5.2最大输出131K → 33K
- Z.ai: GLM 5.2缓存读$0.20 → $0.18-10%
- Z.ai: GLM 5.2输出价$4.10 → $4.00-2%
- Z.ai: GLM 5.2输入价$1.20 → $1.00-17%
- Google: Nano Banana Pro (Gemini 3 Pro Image)上下文窗口66K → 131K
- Google: Nano Banana 2 (Gemini 3.1 Flash Image)最大输出66K → 33K
2026-06-20
- Google: Nano Banana 2 (Gemini 3.1 Flash Image Preview)最大输出66K → 33K
- Arcee AI: Trinity Large Thinking最大输出262K → 80K
- Arcee AI: Trinity Large Thinking输出价$0.85 → $0.80-6%
- Arcee AI: Trinity Large Thinking输入价$0.22 → $0.25+14%
- Z.ai: GLM 5.1最大输出— → 66K
- Z.ai: GLM 5.1缓存读$0.18 → $0.49+169%
- MoonshotAI: Kimi K2.6最大输出262K → 262K
- MoonshotAI: Kimi K2.6缓存读$0.20 → $0.33+65%
- MoonshotAI: Kimi K2.6输入价$0.67 → $0.66-1%
- Tencent: Hy3 preview最大输出262K → —
- Tencent: Hy3 preview缓存读$0.03 → $0.02-28%
- Tencent: Hy3 preview输出价$0.26 → $0.21-19%
- Tencent: Hy3 preview输入价$0.07 → $0.06-5%
2026-06-19
- MoonshotAI: Kimi K2.7 Code缓存读$0.14 → $0.13-10%
- MoonshotAI: Kimi K2.7 Code输出价$3.41 → $3.07-10%
- MoonshotAI: Kimi K2.7 Code输入价$0.68 → $0.61-10%
- MoonshotAI: Kimi K2.7 Code最大输出16K → 262K
- MoonshotAI: Kimi K2.7 Code缓存读$0.15 → $0.14-4%
- MoonshotAI: Kimi K2.7 Code输出价$3.50 → $3.41-3%
- MoonshotAI: Kimi K2.7 Code输入价$0.74 → $0.68-8%
2026-06-18
- OpenAI: gpt-oss-safeguard-20b缓存读$0.04 → $0.04+1%
- Qwen: Qwen3.5-35B-A3B缓存读$0.05 → —
- Qwen: Qwen3.6 35B A3B缓存读$0.05 → —
- Z.ai: GLM 5.2最大输出66K → 131K
- Z.ai: GLM 5.2输出价$3.20 → $4.10+28%
- Qwen: Qwen3.5-35B-A3B最大输出82K → 262K
- MoonshotAI: Kimi K2.6最大输出262K → 262K
- MoonshotAI: Kimi K2.6缓存读$0.34 → $0.20-41%
- MoonshotAI: Kimi K2.6输出价$3.41 → $3.50+3%
- MoonshotAI: Kimi K2.6输入价$0.68 → $0.67-1%
- Qwen: Qwen3.6 35B A3B输入价$0.15 → $0.14-7%
- Z.ai: GLM 5.2最大输出262K → 66K
- Z.ai: GLM 5.2缓存读$0.26 → $0.20-23%
- Z.ai: GLM 5.2输出价$4.40 → $3.20-27%
- Z.ai: GLM 5.2输入价$1.40 → $1.20-14%
2026-06-17
- Google: Nano Banana (Gemini 2.5 Flash Image)最大输出33K → 8K
- Google: Gemini 3 Flash Preview最大输出66K → 66K
2026-06-16
- NVIDIA: Nemotron 3 Ultra缓存读$0.15 → $0.10-33%
- NVIDIA: Nemotron 3 Ultra输出价$2.50 → $2.20-12%
- Z.ai: GLM 4.5 Air最大输出131K → 98K
- Z.ai: GLM 4.5 Air缓存读$0.06 → $0.03-58%
- Z.ai: GLM 4.5 Air输入价$0.13 → $0.13+4%
- Tencent: Hy3 preview最大输出— → 262K
- Tencent: Hy3 preview缓存读$0.02 → $0.03+38%
- Tencent: Hy3 preview输出价$0.21 → $0.26+24%
- Tencent: Hy3 preview输入价$0.06 → $0.07+5%
- DeepSeek: DeepSeek V4 Flash最大输出— → 66K
- DeepSeek: DeepSeek V4 Flash输出价$0.20 → $0.18-8%
- DeepSeek: DeepSeek V4 Flash输入价$0.10 → $0.09-8%
- MoonshotAI: Kimi K2.7 Code最大输出262K → 16K
- MoonshotAI: Kimi K2.7 Code缓存读$0.16 → $0.15-6%
- MoonshotAI: Kimi K2.7 Code输入价$0.75 → $0.74-1%
2026-06-15
- Qwen: Qwen3.5 397B A17B最大输出66K → —
- Qwen: Qwen3.5 397B A17B上下文窗口262K → 256K
- Qwen: Qwen3.5 397B A17B输出价$2.34 → $2.45+5%
- Qwen: Qwen3.5 397B A17B输入价$0.39 → $0.39-1%
- Qwen: Qwen3.5-35B-A3B最大输出262K → 82K
- xAI: Grok 4.20 Multi-Agent输出价$6.00 → $2.50-58%
- xAI: Grok 4.20 Multi-Agent输入价$2.00 → $1.25-37%
- DeepSeek: DeepSeek V4 Flash最大输出66K → —
- DeepSeek: DeepSeek V4 Flash输出价$0.18 → $0.20+9%
- DeepSeek: DeepSeek V4 Flash输入价$0.09 → $0.10+9%
- OpenRouter: Fusion上下文窗口128K → 1M
2026-06-13
- MoonshotAI: Kimi K2.5缓存读$0.09 → —
- DeepSeek: DeepSeek V4 Flash最大输出— → 66K
- DeepSeek: DeepSeek V4 Flash输出价$0.20 → $0.18-8%
- DeepSeek: DeepSeek V4 Flash输入价$0.10 → $0.09-8%
- MoonshotAI: Kimi K2.7 Code最大输出— → 262K
- MoonshotAI: Kimi K2.7 Code缓存读$0.19 → $0.16-16%
- MoonshotAI: Kimi K2.7 Code输出价$4.00 → $3.50-12%
- MoonshotAI: Kimi K2.7 Code输入价$0.95 → $0.75-21%
2026-06-12
- MoonshotAI: Kimi K2.5最大输出262K → —
- MoonshotAI: Kimi K2.5输出价$1.90 → $2.02+7%
- MoonshotAI: Kimi K2.5输入价$0.40 → $0.38-6%
- MoonshotAI: Kimi K2.6最大输出262K → 262K
- MoonshotAI: Kimi K2.6缓存读$0.14 → $0.34+143%
- MoonshotAI: Kimi K2.6输出价$3.39 → $3.41+1%
- MoonshotAI: Kimi K2.6输入价$0.67 → $0.68+1%
- Qwen: Qwen3.6 27B最大输出131K → 262K
- Qwen: Qwen3.6 27B输出价$2.40 → $3.17+32%
- Qwen: Qwen3.6 27B输入价$0.29 → $0.290%
2026-06-11
- 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%
- DeepSeek: DeepSeek V3 0324上下文窗口131K → 164K
- DeepSeek: DeepSeek V4 Flash最大输出131K → —
- DeepSeek: DeepSeek V4 Flash缓存读$0.02 → $0.02+2%
- DeepSeek: DeepSeek V4 Flash输出价$0.20 → $0.200%
- DeepSeek: DeepSeek V4 Flash输入价$0.10 → $0.100%
- DeepSeek: R1 Distill Llama 70B最大输出16K → 8K
- DeepSeek: R1 Distill Llama 70B上下文窗口131K → 128K
- DeepSeek: R1 Distill Llama 70B输入价$0.70 → $0.80+14%
- MiniMax: MiniMax M2.7缓存读$0.05 → $0.05-7%
- MiniMax: MiniMax M2.7输出价$1.08 → $1.00-7%
- MiniMax: MiniMax M2.7输入价$0.27 → $0.25-7%
- Google: Gemma 4 31B最大输出8K → 262K
- Google: Gemma 4 31B输出价$0.36 → $0.35-3%
- MoonshotAI: Kimi K2.6最大输出262K → 262K
- MoonshotAI: Kimi K2.6缓存读$0.34 → $0.14-59%
- MoonshotAI: Kimi K2.6输出价$3.41 → $3.39-1%
- MoonshotAI: Kimi K2.6输入价$0.68 → $0.67-1%
- Qwen: Qwen3.6 35B A3B最大输出262K → 262K
- Qwen: Qwen3.6 35B A3B缓存读— → $0.05
- Qwen: Qwen3.6 35B A3B输入价$0.14 → $0.15+7%
- Qwen: Qwen3.7 Plus缓存写$0.50 → $0.40-20%
- Qwen: Qwen3.7 Plus缓存读$0.08 → $0.06-20%
- Qwen: Qwen3.7 Plus输出价$1.60 → $1.28-20%
- Qwen: Qwen3.7 Plus输入价$0.40 → $0.32-20%
2026-06-10
- DeepSeek: DeepSeek V3 0324上下文窗口164K → 131K
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。