We've evaluated Gemini 4 Carbon and Gemini 4 Argon on FrontierCode 2.0, our coding benchmark that me...
TL;DR · AI 摘要
Cognition在FrontierCode 2.0基准测试中显示Gemini 4 Carbon和Argon的性能与成本表现，但落后于部分竞品模型。
核心要点
- Gemini 4 Carbon得分为40%，单任务成本$4.04，仅次于Claude Opus 5.5等模型
- Gemini 4 Argon得分为26.3%，成本$6.41，领先Muse Wonder 1.0但落后Claude Haiku 5.5
- FrontierCode 2.0基准测试衡量模型超越规格书要求的编程能力
结构提纲
按章节快速跳转。
- §测试背景
介绍FrontierCode 2.0基准测试的目标与评估维度。
报告其40%得分与$4.04成本，定位行业顶尖但非最优。
披露26.3%得分与$6.41成本，处于中游位置。
- ›竞品对比
与Claude Opus 5.5、GPT-6 Astra等模型进行横向性能与成本对比。
思维导图
用一张图看清主题之间的关系。
查看大纲文本（无障碍 / 无 JS 友好）
- FrontierCode 2.0模型评估
- 测试模型
- Gemini 4 Carbon
- Gemini 4 Argon
- 性能指标
- 得分40%/$4.04
- 得分26.3%/$6.41
- 竞品对比
- Claude Opus 5.5领先
- GPT-6 Astra领先
金句 / Highlights
值得收藏与分享的关键句。
Gemini 4 Carbon以40%得分接近SOTA，但单任务成本$4.04高于Claude Opus 5.5
Gemini 4 Argon在26.3%得分段落中成本$6.41，超越Muse Wonder 1.0但落后Claude Haiku 5.5
FrontierCode 2.0基准测试专门评估模型超越规格书要求的编程能力
Cognition on X: "We've evaluated Gemini 4 Carbon and Gemini 4 Argon on FrontierCode 2.0, our coding benchmark that measure's a model's ability to go beyond the spec. Gemini 4 Carbon scores near SOTA with a score of 40% and a cost of $4.04 per task, behind Claude Opus 5.5, GPT-6 Astra, and Claude S… / X
@cognition
We've evaluated Gemini 4 Carbon and Gemini 4 Argon on FrontierCode 2.0, our coding benchmark that measure's a model's ability to go beyond the spec. Gemini 4 Carbon scores near SOTA with a score of 40% and a cost of $4.04 per task, behind Claude Opus 5.5, GPT-6 Astra, and Claude Sonnet 5.5. Gemini 4 Argon scores 26.3% with a cost of $6.41 per task, ahead of Muse Wonder 1.0, but behind Claude Haiku 5.5 and Grok 4.7. See the leaderboard with full scores:
cognition.com/frontiercode
1:45 AM · Oct 30, 2026
·
242K
Views
149
119
1.1K
210