
Gemini 4 Carbon 对比 DeepSeek V4 Pro:37 个指数点与 4.4 倍的每任务成本,究竟换来了什么
- Anthropic新Anthropic: Claude Fable 5.22026-11-09$8.00 / $40.00 每百万 tokens · 100 tok/s
- orca新Orca: OrcaC2 1.02026-11-07235 tok/s
- Orca新OrcaCyber: GLM 5.3 Abliterated / Uncensored2026-11-04$3.00 / $5.00 每百万 tokens · 119 tok/s
- Orca新OrcaCyber: Hy4 Abliterated / Uncensored2026-11-04$0.80 / $1.60 每百万 tokens · 220 tok/s
- Orca新OrcaCyber: Kimi K3.1 Pro Abliterated / Uncensored2026-11-04$4.00 / $20.00 每百万 tokens · 39 tok/s
- Orca新OrcaCyber: MiniMax M3.1 Flash Abliterated / Uncensored2026-10-29$0.80 / $1.60 每百万 tokens · 122 tok/s
- OrcaOrcaCyber: GLM 5.3 Flash Abliterated / Uncensored2026-10-27$0.80 / $1.60 每百万 tokens · 59 tok/s
- OrcaOrcaCyber: Qwen3.8 Flash Next Abliterated / Uncensored2026-10-26$0.80 / $1.60 每百万 tokens · 96 tok/s
- OrcaOrcaCyber: MiMo V2.6 Flash Abliterated / Uncensored2026-10-26$0.80 / $1.60 每百万 tokens · 40 tok/s
- MoonshotMoonshotAI: Kimi K3.1 Pro2026-10-2447智能
- OrcaOrcaCyber: DeepSeek V4.1 Flash Abliterated / Uncensored2026-10-23$0.80 / $1.60 每百万 tokens · 114 tok/s
- minimaxMiniMax: M3.1 Flash2026-10-1929智能
- OrcaOrca: OrcaCyber Ultra 1.02026-10-17$5.00 / $10.00 每百万 tokens · 103 tok/s
- deepseekDeepSeek: DeepSeek V4.1 Flash 10172026-10-1733智能
- DeepSeekDeepSeek: DeepSeek V4.1 Flash2026-10-1740智能
- OrcaOrca: OrcaCyber Zero 1.52026-10-10$3.00 / $7.50 每百万 tokens · 41 tok/s
- openaiOpenAI: GPT-6.1 Sol2026-09-2955智能
- anthropicAnthropic: Claude Sonnet 5.52026-09-2855智能
- typesafeTypeSafe: Jev 1.132026-09-24$0.04 / $0.00 每百万 tokens · 508 tok/s
- OpenAIOpenAI: GPT-6 Luna2026-09-2228智能
Put Gemini 4 Carbon and DeepSeek V4 Pro on the same scale and the result is not close: 60 against 23 on Artificial Analysis Intelligence Index v5.0.1, a 37-point gap between the fourth-best model on the board and one sitting below the median. Put them on the cost scale and it is not close either, in the other direction — $7.76 per index task against $1.77. The interesting question is not which is better. It is which of those two numbers you are actually paying for in the work you have, because for a large share of production traffic the 37 points go unused while the 4.4x bill arrives every month.
Both models are callable today, which makes this a real comparison rather than a roadmap exercise. Gemini 4 Carbon opened to paid API developers on 10 November 2026 after a twelve-day gated rollout. DeepSeek V4 Pro has been generally available since 13 August 2026 as the 0813 checkpoint, with open weights under MIT.
基于相反假设的两种模型
规格说明在任何基准测试之前就解释了大部分行为差异。
• Price, standard rate — Gemini 4 Carbon: $2 in / $10 out per million until 30 November 2026, then $4 / $20. DeepSeek V4 Pro: $0.66 / $1.98 off-peak and $1.32 / $3.96 at peak, straight from DeepSeek's own rate card.
• Cached input — Carbon $0.10 per million now, $0.20 after 1 December, implicit caching only. DeepSeek $0.022 off-peak and $0.044 at peak, a 97% discount on the fresh-token rate.
• 上下文与输出——二者均具有 100 万 token 的窗口。Carbon 将同步输出上限设为 65,536 个 token,并在长解码时允许最多 1,048,576 个;V4 Pro 的上限为 384,000 个。
• Input modality — Carbon takes text, image, audio, video and PDF. V4 Pro is text only, and DeepSeek's own pricing page says vision is not supported on v4-pro.
• 权重 — Carbon 是闭源的。V4 Pro 以 MIT 许可证发布,采用 1.6 万亿参数的混合专家架构,拥有 490 亿活跃参数,因此你可以在许可证条款约束下自行托管、微调和再分发。
• 推理强度控制——Carbon 提供低、中、高三档思考强度。V4 Pro 提供低、高、最高三档推理调节。
模态这条线并不是一个无关紧要的脚注。如果你的流水线需要接收截图、扫描版 PDF 或通话录音,那么无论价格如何,V4 Pro 都不能成为该技术栈环节中的模型,比较也就坍缩为单一候选。如果你的流量是代码、日志和文本文档,那么这一差距就确实是一个选择问题。
分解后的 37 个要点
那么大的综合分差,通常只是靠少数几项评测撑起来的,而独立榜单对这两个模型显示的正是这种情况。Artificial Analysis 是在高推理强度下测 Carbon、在最高推理强度下测 V4 Pro 0813 的;这两个数字都出自该评测方,而不是厂商。
• 每项智能指数任务的成本——Carbon 为 7.76 美元,V4 Pro 为 1.77 美元。该比值已将冗长程度纳入考量,因此是两者之间最公平的单一经济指标。
• 在该指数上消耗的输出 token:Carbon 2.05 亿,V4 Pro 1.19 亿,而中位数为 7280 万。两个模型的输出都偏冗长;Carbon 大约是其 1.7 倍,价格差距中有相当一部分正来自这里,而不是仅来自每 token 费率。
• 速度与延迟——Carbon 80 的输出速度为每秒 80 个 token,生成首个回答 token 耗时 66.85 秒;V4 Pro 为每秒 67 个 token,耗时 67.71 秒。在这一维度上,两者几乎完全相同,而且都足够慢:如果不采用流式传输并保持耐心,两者都不适合用于交互式聊天路径。
推理力度敏感度——Carbon 在高档得 60 分,中档 58 分,低档 54 分。V4 Pro 的非推理配置得 13 分,而在最高档为 23 分。把推理力度调低时,更便宜的模型损失要大得多,因此“便宜地运行”对双方而言都不是无代价的开关。
On the vendor side the numbers are loud and unaudited, and they do not overlap enough to compare directly. DeepSeek's own card for the 0813 build reports Terminal-Bench 2.1 at 87.9, DeepSWE at 62.7, CyberGym at 83.3, SWE-bench Verified at 80.6 and a Codeforces rating of 3206 — all vendor-reported, none reproduced by a neutral party at the time of writing. Google's card for Carbon reports SEC-bench Pro at 88.2%, ExploitGym at 54.8% and CWE-bench at 71.0%. Note what happens if you try to line those up: CyberGym and ExploitGym are different harnesses, Terminal-Bench 2.1 and Terminal-Bench 4.0 are different versions of a moving target, and only one of the two vendors published a cyber score on a benchmark the other also ran. Any chart setting 83.3 beside 54.8 is comparing a scale to a different scale.

没有人为其做预算的计费时钟
DeepSeek's price is not one price, and this is the most under-reported difference between the two models. V4 Pro bills at half rate outside a defined peak window: peak is 01:00–04:00 and 06:00–10:00 UTC, Monday to Friday, excluding Chinese public holidays. Everything else — nights, weekends, holidays — is off-peak.
一周五个工作日、每天 24 小时中有七小时,约占日历周的 21%。如果你的批处理任务由 UTC 时区的 cron 调度,这很容易对齐。如果你的流量来自一个有美国或欧洲白天曲线的在线产品,那么其中相当大一部分会落在峰值窗口内,你实际混合后的费率就在 0.66 美元到 1.32 美元之间,而不是营销页面上写的那个数字。
Carbon 的计费时钟走向相反。它在 11 月 30 日之前保持不变,然后翻倍。两种模型,账单波动的两个完全不同来源:一个随一天中的小时变化,另一个随日历日期变化。两者都无法从单个每 token 数字中看出来,而将 $4/$20 与 $0.66/$1.98 相比较的对比,描述的是一侧的最坏情况和另一侧的最好情况。
坦白说:按 Carbon 的 introductory 价格对比 V4 Pro 的峰值价格,Carbon 在输入上是 1.5 倍,在输出上是 2.5 倍。12 月 1 日之后,与 off-peak 相比,则是 6 倍和 10 倍。每任务成本 4.4 倍这个数字介于两者之间,才是值得用来做规划的数字,因为它考虑了每个模型实际完成任务需要多少 token。

差距值得付出代价的地方
与其给出定论,不如给出四种经得起这两个模型实际形态检验的划分。
Cyber and vulnerability work. This is Carbon's home ground and the reason Google gated it to defenders first. Carbon reads code, images and PDFs in one call, which is what a vulnerability triage pipeline needs, and Google's own reported numbers put it at the top of the cyber boards it chose to publish. V4 Pro has respectable self-reported cyber scores and cannot see a screenshot. For a security team, Carbon at 6x the input price is not obviously the wrong call.
大批量分类、提取与路由。 把这一项交给 V4 Pro。37 点的综合差距几乎说明不了一个模型能否给支持工单打标签,或者从发票中提取字段;而在每百万输入 0.66 美元的价格下,便宜模型在算术上早就赢了,远没到在质量上输掉的地步。大多数团队在这方面都买高配了。
长上下文文档处理。两个窗口均为 100 万 token,因此原始容量持平。差异在于模态和输出上限:Carbon 可直接读取扫描版 PDF,并可输出最多 100 万 token;V4 Pro 则需要文本已提取完成,且输出上限为 38.4 万 token。如果你的流程已经完成 OCR,V4 Pro 同样具备 100 万 token 的上下文窗口,而输出价格仅为其十分之一。
任何你可能不得不自行托管的内容。唯一的选择是 V4 Pro。MIT 权重意味着没有供应商能够撤回该模型、重新定价,或将其置于某个计划的门槛之后——而在 Carbon 的访问条件于十四天内两次变化、其价格又将于 12 月 1 日再次变化的这一年,这并非假设性的担忧。
路由答案,并附上诚实的提醒
这个决策的成熟做法不是二选一,而是让 V4 Pro 处理大部分流量,并把困难案例升级到前沿模型——这正是我们的路由 DSL 存在的组合方式:一个处理流量的低成本分支、一个在低置信度或验证失败时触发的升级条件,以及一个位于两者之前的 OpenAI 兼容端点。
What is on that endpoint today is deepseek/deepseek-v4-pro at DeepSeek's own $0.66/$1.98, alongside google/gemini-4-argon at $2/$10 — Argon being the Gemini 4 model that carries the same 1M window and a 50 on the same index, at half Carbon's post-December output price. Gemini 4 Carbon is not in our catalogue; it is reached through Google's paid API, and a Carbon branch in a pipeline is a direct integration. Because provider list price is passed through with 0% markup across the 200-plus models on the key, a vendor rate change lands the day it happens, which for Carbon means 1 December.
自动故障转移才是改变廉价模型成本权衡的关键所在。团队之所以过度购买前沿模型容量,是因为担心在真正重要的那 5% 请求上得到糟糕的答案。故障转移和升级机制让廉价模型承担它能处理好的 95% 请求,而当请求超出其能力范围时,可以通过同一个端点调用昂贵模型——而不是为每一个 token 都支付前沿模型的价格,去防范本可通过路由化解的尾部风险。

什么会改变这个推荐
Two developments, both plausible within a quarter. If a neutral evaluator runs Carbon on a cyber benchmark that DeepSeek also published, the two scorecards become comparable for the first time and the gap stops being an inference. And if DeepSeek ships the V4.1-Pro that its own migration note pointed at, the cheap side of this comparison gets stronger without the expensive side moving — which would make the escalation split even more attractive than it is now.
在此之前,站得住脚的立场并不光鲜:V4 Pro 用于走量,Carbon 只用在多模态输入或网络安全能力真正起关键作用的地方,而且不要在这个二十天内就翻倍的价格上、在买卖任何一方签下长期合同。
