Gemini 3.8 Flash vs Tencent Hy4

vs

什么时候选哪个

基础档每百万输出 token 的价格为 $3.75 和 $2.501(gemini-3.8-flash 对 hy4)。 单次请求要结合图片和音频或视频时选 gemini-3.8-flash。 与 gemini-3.8-flash 相比,中英文长文档夹有截图时,可评估 hy4。

Benchmark 成绩

Tencent Hy4:供应商没有公布 benchmark 成绩。

高于同类均值无更高分Gemini 3.8 Flash12 / 175 / 17
Gemini 3.8 Flash Tencent Hy4 其他参测模型 同类均值 ★ 没有模型得分更高
DeepSWE 1.1
73.7%
N/A
BioMysteryBench hard
没有模型得分更高 56.5%
N/A
OSWorld 2.0 Partial score, batch tool enabled
59%
N/A
HealthBench Professional
52.1%
N/A
Finance Agent v2
没有模型得分更高 61.4%
N/A
Legal Agent Benchmark
10%
N/A
GPQA Diamond
95.3%
N/A
CharXiv (RQ) no tools
没有模型得分更高 86.2%
N/A

供应商公布: Alibaba (Qwen) Alibaba (Qwen) Alibaba (Qwen) Alibaba (Qwen) Alibaba (Qwen) Alibaba (Qwen) Anthropic Anthropic Anthropic Anthropic ByteDance DeepSeek DeepSeek Google Google Google Google MiniMax Moonshot OpenAI OpenAI OpenAI OpenAI OpenAI OpenAI OpenAI Tencent Z.ai Z.ai Z.ai Z.ai

价格

Gemini 3.8 Flash Tencent Hy4 Δ
输入 / 1M token $0.75 $0.834 0.9×
输出 / 1M token $3.75 $2.501 1.5×
缓存读取 / 1M token $0.075 $0.042 1.8×

价格取自构建时的实时目录,最新价格见各模型页面。

两个模型所处的位置:全部 86 个按同一单位计费的聊天模型的每 1M token 输入价分布(对数刻度)

Gemini 3.8 Flash · $0.75 Tencent Hy4 · $0.834
$0.05 · Qwen3 VL Flash $30 · GPT-5.4 Pro

能力

Gemini 3.8 Flash Tencent Hy4
工具调用 是 是
思考控制 是,但供应商未公布调节参数 是,但供应商未公布调节参数
结构化输出 是 -
提示词缓存 隐式 + 显式 隐式(自动)
缓存有效期 未公布 未公布
最小缓存前缀 4,096 个 token 未公布

规格

Gemini 3.8 Flash Tencent Hy4
输入模态 文本 图像 音频 视频 文本 图像
输出模态 文本 文本
发布日期 2026-09-02 2026-11-04
知识截止日期 2026-03 -
上下文窗口 1,048,576 1,048,576
最大输出 65,536 65,536

规格照录自各供应商的文档;供应商没有公布的项目,对应的行直接省略,不做推断。 完整来源: Gemini 3.8 Flash · Tencent Hy4

同一条提示词,两个模型,经网关实测

提示词 Explain to a CFO, in exactly three sentences, why our LLM bill doubled after we enabled reasoning. 检查 恰好 3 句

Gemini 3.8 Flash 通过 · 3 sentences

When we enabled reasoning, the model began generating hidden "thinking tokens" to work through logic step-by-step before producing a final answer. Because AI vendors bill for every single token processed—visible or not—this internal deliberation dramatically inflated our billable volume per query. In short, while our total number of user requests remained flat, the cost per transaction doubled to buy higher accuracy on complex tasks.

输出 705 tok (+624 思考) 延迟 6.4 s

Tencent Hy4 通过 · 3 sentences

Enabling reasoning changed the unit economics of each request: the model now produces extra “reasoning tokens” and consumes more compute before it returns an answer, and those tokens are billed even if users never see them. Because our pricing is largely token-based, that extra work roughly doubled average cost per call, so flat request volume could still produce a doubled invoice. We are paying for better accuracy and less downstream rework on complex tasks, so the spend is justified only in high-value workflows and should be managed by cost per successful outcome.

输出 1519 tok (+1409 思考) 延迟 21.0 s

指令遵循(恰好三句,数得出来)、受众适配(对 CFO 说话的口吻),以及下方 token 计数暴露出的隐藏思考计费差额。

提示词 This function is supposed to return the longest run of consecutive equal items, but callers report it is off by one on single-element inputs and misses runs that end at the last index. Fix it and explain the bug in one paragraph. 检查 修复通过测试

Gemini 3.8 Flash 通过 · 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 best = max(best, cur) else: cur = 1 return best ``` ### Explanation of the Bug The original function counted equal-neighbor transitions rather than the actual items making up each run. Because `cur` started at `0` (and was reset to `0` on mismatches), every run was undercounted by one: single-element lists returned `0` because the loop never executed, and any subsequent run—including those ending at the last index—recorded a count of $N - 1$ instead of $N$. The fix handles empty input upfront, initializes both `cur` and `best` to `1` (since any single element forms a run of length 1), and resets `cur` to `1` whenever a new run begins.

输出 1393 tok (+1149 思考) 延迟 10.1 s

Tencent Hy4 通过 · 8/8 cases

```python def longest_run(xs): n = len(xs) if n == 0: return 0 best = cur = 1 for i in range(1, n): if xs[i] == xs[i - 1]: cur += 1 else: cur = 1 best = max(best, cur) return best ``` The loop range already includes the last index; the real bug is that `cur` and `best` start at `0` and only increase on equal adjacent comparisons, so the function counts equality links (`L - 1` for a run of `L` equal items), not the items in the run: a single item has no link, and the first item of a final run is never counted, with no end-of-list flush to compensate. […]

输出 4509 tok (+4263 思考) 延迟 56.0 s

修复是否真的正确(能运行)、解释的信息密度,以及在一个范围明确的任务上的 token 效率。

提示词 Extract a JSON object with fields {company, quarter, revenue_usd_m, growth_pct, guidance} from this text. Use null for anything not stated; add no other fields. 检查 合法 JSON,严格符合 schema

Gemini 3.8 Flash 通过 · 5/5 fields, guidance null

```json { "company": "Meridian Robotics", "quarter": "September quarter", "revenue_usd_m": 412, "growth_pct": null, "guidance": null } ```

输出 3989 tok (+3932 思考) 延迟 30.4 s

Tencent Hy4 通过 · 5/5 fields, guidance "withheld pending the Q4 close"

{"company":"Meridian Robotics","quarter":"September quarter","revenue_usd_m":412,"growth_pct":"eleven and change","guidance":"withheld pending the Q4 close"}

输出 9869 tok (+9827 思考) 延迟 125.2 s

是否严守 schema(不臆造字段)、能否顶住幻觉压力(原文明说暂不给出 guidance),以及结构化输出路径的差异。

提示词 Write a 120-word product announcement for an API gateway feature that caches prompts across providers. Forbidden words: "seamless", "unlock", "game-changer", "revolutionize", "empower". Exactly one sentence must be a question. 检查 120 词,0 个禁用词

Gemini 3.8 Flash 通过 · 120 words, 0 banned, 1 question

Why pay multiple model providers for the exact same output? Introducing Universal Prompt Cache, our latest API gateway capability engineered to cut compute expenses and drop inference latency. When your application makes a call, the gateway inspects a central memory layer before routing traffic to external LLMs. If an identical query was previously processed by OpenAI, Anthropic, or Mistral, our gateway returns that response immediately. This shared cache eliminates duplicate token fees and insulates your production apps from vendor rate limits or unexpected downtime. Developers can easily customize expiration settings, enforce strict data privacy controls, and configure invalidation logic across every endpoint. Stop wasting your budget on repeated queries. Enable prompt caching in your dashboard to accelerate your pipeline today.

输出 3833 tok (+3688 思考) 延迟 21.8 s

Tencent Hy4 通过 · 120 words, 0 banned, 1 question

Today we are introducing Cross Provider Prompt Cache for our API gateway, a feature that caches prompts and securely reuses them across all connected model providers. Teams can cut duplicate calls, reduce latency, and keep spend predictable when identical instructions appear in chat, search, agents, or batch jobs. What if one cached prompt could support traffic routed to any provider in your stack? Gateway policies control cache keys, time to live, tenant isolation, redaction, and fallback behavior, while usage metrics show hit rates and savings by application. Setup requires one configuration change, and existing client code stays intact. Cross Provider Prompt Cache is available in private beta today for enterprise plans, with general availability next month across all supported regions.

输出 5744 tok (+5601 思考) 延迟 57.6 s

是否守住约束(字数预算、禁用词表、只能有一个问句)、文风特征,以及长度控制。

改一行代码就能在两个模型之间切换

下方每个标签页里都有两个模型 ID,高亮的那两行是唯一要改的地方。端点不变,API key 不变,请求结构也不变。

from openai import OpenAI

client = OpenAI(
    base_url="https://synthorai.io/v1",
    api_key="sk-syn-...",
)

resp = client.chat.completions.create(
    model="gemini-3.8-flash",
    # model="hy4",  # 取消注释此行,注释上一行
    messages=[{"role": "user", "content": "Summarize this diff"}],
    reasoning_effort="medium",
)
print(resp.choices[0].message.content)

获取 API key →

常见问题

Gemini 3.8 Flash 和 Tencent Hy4 哪个更便宜?

按「输入 / 1M token」算,Gemini 3.8 Flash 更便宜($0.75 对 $0.834,相差 1.1×)。其他计费项的结论可能相反,完整价格见上方表格,实际成本取决于你的用量构成。

不用分别集成两次,就能对 Gemini 3.8 Flash 和 Tencent Hy4 做 A/B 测试吗?

可以。两个模型走同一个 OpenAI 兼容端点,用同一个 API key,切换时只要改一行里的模型名,所以可以给两个模型各分一部分流量,直接对比账单。

Gemini 3.8 Flash 和 Tencent Hy4 支持提示词缓存吗?

支持。两个模型的缓存读取价都低于各自的输入价,所以前缀能反复命中缓存的负载,实际成本会比按官网价估算的低。具体的缓存读取价见上方价格表。

相关对比

我们的实测研究