Gemini 3.1 Flash-Lite vs Gemini 3.7 Flash
什麼情況選哪一個
這兩款模型均具備 1048576-token 的脈絡視窗與 65536-token 的最大輸出限制，且支援相同的文字、圖片、影片與音訊輸入以及文字輸出，因此差異在於價格與推論能力：gemini-3.1-flash-lite-preview 每百萬輸入收費 $0.25、輸出收費 $1.5，而 gemini-3.7-flash 的輸入收費為其 3x ($0.75)、輸出收費為 2.5x ($3.75)，並新增了推論能力旗標。對於成本為首要考量的大量多模態工作，請選擇 gemini-3.1-flash-lite-preview；若任務需要推論能力，或是有重複的提示詞可適用其 $0.075 的快取讀取費率時，請選擇 gemini-3.7-flash。
Benchmark 成績
供應商公布： Alibaba (Qwen) Anthropic ByteDance DeepSeek Google Moonshot OpenAI Tencent Z.ai
定價
| Gemini 3.1 Flash-Lite | Gemini 3.7 Flash | Δ | |
|---|---|---|---|
| 輸入 / 1M tokens | $0.25 | $0.75 | 0.33× |
| 輸出 / 1M tokens | $1.5 | $3.75 | 0.4× |
| 快取讀取 / 1M tokens | - | $0.075 | - |
費率取自網站建置時的即時目錄；各模型頁面都列有最新的價目。
兩者的相對位置：每 1M tokens 的輸入價格，涵蓋同一計費單位下全部 76 個聊天模型（對數尺度）
功能
| Gemini 3.1 Flash-Lite | Gemini 3.7 Flash | |
|---|---|---|
| 工具使用 | 是 | 是 |
| 思考控制 | 是，但供應商未公布調整參數 | 是，但供應商未公布調整參數 |
| 結構化輸出 | 是 | 是 |
| 提示詞快取 | 隱式 + 顯式 | 隱式 + 顯式 |
| 快取存活時間 | 未公布 | 未公布 |
| 最小快取前綴 | 4096 個 token | 4096 個 token |
規格
| Gemini 3.1 Flash-Lite | Gemini 3.7 Flash | |
|---|---|---|
| 輸入模態 | 文字 圖像 音訊 影片 | 文字 圖像 音訊 影片 |
| 輸出模態 | 文字 | 文字 |
| 發布日期 | 2026-03-03 | 2026-08-13 |
| 知識截止日期 | 2025-01 | 2026-03 |
| 上下文視窗 | 1M | 1M |
| 最大輸出 | 66K | 66K |
規格摘錄自各供應商的文件；供應商沒有公布的項目就直接略過，不自行推測。 完整來源： Gemini 3.1 Flash-Lite · Gemini 3.7 Flash
同一段提示詞，兩個模型，經閘道實測
Gemini 3.1 Flash-Lite 通過 · 3 sentences
Enabling reasoning models increases your costs because the system now generates a significantly higher volume of "hidden" chain-of-thought tokens required to process complex logic. Since you are billed for every token produced during this internal deliberation phase, your total output volume has expanded alongside the depth of the analysis. You are essentially paying for the model to "think out loud" before arriving at a final answer, which directly scales your operational expenditure.
輸出 86 tok 延遲 1.4 s
Gemini 3.7 Flash 通過 · 3 sentences
Enabling reasoning causes the model to generate a high volume of invisible "thinking tokens" behind the scenes to deliberate and solve complex logic before delivering a final response. Because our vendor bills strictly on token volume, this additional internal computation effectively doubled our billable output per query. In short, our costs increased because we are now paying for the extra compute capacity required to produce significantly higher-accuracy answers.
輸出 878 tok (+799 思考) 延遲 6.9 s
是否照指令做（剛好三句，數得出來）、語氣是否貼合對象（寫給 CFO 看），以及下方 token 計量顯示出的隱藏思考計費落差。
Gemini 3.1 Flash-Lite 通過 · 8/8 cases
### Fixed Code ```python def longest_run(xs): if not xs: return 0 best = 1 cur = 1 for i in range(1, len(xs)): if xs[i] == xs[i - 1]: cur += 1 best = max(best, cur) else: cur = 1 return best ``` ### Explanation of the Bug The original function failed because it initialized `best` and `cur` to `0` and effectively counted "transitions" (how many times an element matched its predecessor) rather than the total count of elements in a run. By starting at `0`, a single-element list returned `0` instead of `1`, and because `cur` reset to `0` and only incremented when a match was found, the function effectively counted "extra" matches beyond the first element, failing to account for the initial element of any run. […]
輸出 271 tok 延遲 2.0 s
Gemini 3.7 Flash 通過 · 8/8 cases
Here is the corrected function: ```python def longest_run(xs): if not xs: return 0 best = 1 cur = 1 for i in range(1, len(xs)): if xs[i] == xs[i - 1]: cur += 1 else: cur = 1 best = max(best, cur) return best ``` ### Explanation of the Bug The original function tracked equal adjacent transitions instead of the actual number of elements in a run, causing all non-empty run lengths to be undercounted by one (e.g., returning `0` instead of `1` for single-element lists because the loop never executes, and `1` instead of `2` for pairs). Initializing both `best` and `cur` to `1` (after handling empty inputs) and resetting `cur` to `1` on a mismatch ensures every individual element is counted as a valid run of at least length one.
輸出 1050 tok (+827 思考) 延遲 6.4 s
修正是否真的正確（跑得起來）、說明的資訊密度，以及在範圍明確的任務上的 token 效率。
Gemini 3.1 Flash-Lite 通過 · 5/5 fields, guidance null
```json { "company": "Meridian Robotics", "quarter": "September", "revenue_usd_m": 412, "growth_pct": null, "guidance": null } ```
輸出 56 tok 延遲 1.3 s
Gemini 3.7 Flash 通過 · 5/5 fields, guidance "withheld"
```json { "company": "Meridian Robotics", "quarter": "September quarter", "revenue_usd_m": 412, "growth_pct": null, "guidance": "withheld" } ```
輸出 918 tok (+858 思考) 延遲 6.2 s
是否遵守 schema（不自創欄位）、面對幻覺誘因的表現（題目明確不給 guidance），以及結構化輸出路徑的差異。
Gemini 3.1 Flash-Lite 通過 · 125 words, 0 banned, 1 question
We are launching a new caching layer for our API gateway that stores LLM responses across multiple providers. By capturing frequently requested prompts, this feature significantly lowers latency and reduces operational costs for your applications. Instead of querying expensive models for repetitive tasks, your system retrieves identical outputs instantly from our high-speed cache. This integration maintains consistency while supporting diverse provider workflows, ensuring your infrastructure remains both efficient and scalable. Are you ready to optimize your token usage and improve response times for every user? Configuration takes only minutes through our existing dashboard. This addition provides a practical strategy to manage API spend without sacrificing performance or quality. […]
輸出 140 tok 延遲 2.3 s
Gemini 3.7 Flash 通過 · 120 words, 0 banned, 1 question
Why pay twice for identical AI queries simply because you routed them to different model vendors? Today, we introduce Universal Prompt Caching directly within our unified API gateway architecture. This capability stores repeated prompt contexts across OpenAI, Anthropic, and local models, instantly returning stored results to eliminate redundant computation fees. When your application sends an LLM request, the gateway inspects the payload, identifies semantic matches, and returns accurate cached responses in under ten milliseconds. Engineering teams can now slash inference latency by eighty percent while dramatically reducing monthly token expenditures across diverse production deployments. You retain complete privacy control, flexible cache eviction policies, and granular metrics through a single dashboard. Update your routing settings today to accelerate overall system performance.
輸出 2858 tok (+2718 思考) 延遲 14.1 s
是否遵守限制（字數上限、禁用詞清單、只能有一句問句）、文字風格的特徵，以及長度控制。
改一行程式碼就能在兩者之間切換
下面每個頁籤都列了這兩個模型 ID，要改的只有醒目標示的那兩行。端點、金鑰和請求格式都不變。
from openai import OpenAI
client = OpenAI(
base_url="https://synthorai.io/v1",
api_key="sk-syn-...",
)
resp = client.chat.completions.create(
model="gemini-3.1-flash-lite-preview",
# model="gemini-3.7-flash", # 取消這一行的註解，並把上一行註解掉
messages=[{"role": "user", "content": "Summarize this diff"}],
)
print(resp.choices[0].message.content)import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://synthorai.io/v1",
apiKey: "sk-syn-...",
});
const resp = await client.chat.completions.create({
model: "gemini-3.1-flash-lite-preview",
// model: "gemini-3.7-flash", // 取消這一行的註解，並把上一行註解掉
messages: [{ role: "user", content: "Summarize this diff" }],
});
console.log(resp.choices[0].message.content);curl https://synthorai.io/v1/chat/completions \
-H "Authorization: Bearer sk-syn-..." \
-H "Content-Type: application/json" \
-d '{
"model": "gemini-3.1-flash-lite-preview",
# "model": "gemini-3.7-flash", # 取消這一行的註解，並把上一行註解掉
"messages": [{"role": "user", "content": "Hello"}]
}'package main
import (
"context"
"fmt"
"github.com/openai/openai-go/v3"
"github.com/openai/openai-go/v3/option"
)
func main() {
client := openai.NewClient(
option.WithBaseURL("https://synthorai.io/v1"),
option.WithAPIKey("sk-syn-..."),
)
resp, _ := client.Chat.Completions.New(context.TODO(), openai.ChatCompletionNewParams{
Model: "gemini-3.1-flash-lite-preview",
// Model: "gemini-3.7-flash", // 取消這一行的註解，並把上一行註解掉
Messages: []openai.ChatCompletionMessageParamUnion{
openai.UserMessage("Summarize this diff"),
},
})
fmt.Println(resp.Choices[0].Message.Content)
}import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.chat.completions.*;
OpenAIClient client = OpenAIOkHttpClient.builder()
.baseUrl("https://synthorai.io/v1")
.apiKey("sk-syn-...")
.build();
ChatCompletion resp = client.chat().completions().create(
ChatCompletionCreateParams.builder()
.model("gemini-3.1-flash-lite-preview")
// .model("gemini-3.7-flash") // 取消這一行的註解，並把上一行註解掉
.addUserMessage("Summarize this diff")
.build());
System.out.println(resp.choices().get(0).message().content().orElse(""));常見問題
Gemini 3.1 Flash-Lite 和 Gemini 3.7 Flash 哪個比較便宜？
以「輸入 / 1M tokens」來看，Gemini 3.1 Flash-Lite 比較便宜（$0.25 對 $0.75，相差 3.0×）。其他項目的結果可能相反，完整價目請看上表；實際成本要看你的用量組合。
可以只串接一次，就對 Gemini 3.1 Flash-Lite 和 Gemini 3.7 Flash 做 A/B 測試嗎？
可以。兩個模型都走同一個 OpenAI 相容端點，用的也是同一把 API 金鑰，切換時只要改一行裡的模型名稱字串。你可以把一部分流量分別導到兩邊，再直接比較帳單。
Gemini 3.1 Flash-Lite 與 Gemini 3.7 Flash 支援提示詞快取嗎？
我們的資料來源只列出其中一個模型的快取讀取價格；沒有列出費率的那一個，代表它的供應商沒有為快取讀取另訂價格。