GPT-5.5
PulseAugur coverage of GPT-5.5 — every cluster mentioning GPT-5.5 across labs, papers, and developer communities, ranked by signal.
- developed by GPT-5.5-Instant 95%
- competes with DeepSeek 90%
- developed GPT-5.4 90%
- used by DeepSeek V4-Flash 90%
- used by Trusted Access for Cyber 90%
- developed by GPT-5.1 90%
- developed by Romain Huet 90%
- instance of Ethan Mollick 90%
- competes with Grok 4.3 85%
- competes with Mythos 80%
- competes with Claude Code 80%
- competes with Gemini 2.5-Flash 80%
- 2026-05-17 product_launch OpenAI released GPT-5.5, a new iteration of its language model.
- 2026-05-17 product_launch OpenAI designates GPT-5.5 as the primary upgrade path for older models.
- 2026-05-14 product_launch OpenAI has released its new model, GPT-5.5, via API. 来源
- 2026-05-14 research_milestone GPT-5.5 and Claude Mythos showed comparable performance in vulnerability-finding tasks during a UK AI Security Institute evaluation.
- 2026-05-12 product_launch OpenAI's GPT-5.5 launch has led to a surge in user adoption and revenue.
- 2026-05-11 product_launch OpenAI has doubled the list price for its GPT-5.5 model, leading to higher real-world costs for developers.
- 2026-05-11 product_launch OpenAI launched the GPT-5.5 model with significant price increases.
- 2026-05-10 research_milestone GPT-5.5 achieved a higher score than Claude Opus on the Artificial Analysis intelligence benchmark. 来源
- 2026-05-10 product_launch OpenAI launched GPT-5.5 with a significant price increase over its predecessor.
- 2026-05-08 product_launch GPT-5.5 launched with a significant price increase compared to its predecessor.
- 2026-05-07 product_launch OpenAI launched GPT-5.5 with a significant price increase over its predecessor.
- 2026-04-30 product_launch OpenAI released its new GPT-5.5 model, showing competitiveness with leading models.
- 2026-04-30 product_launch OpenAI released its new GPT-5.5 model.
- 2026-04-23 product_launch OpenAI launched its GPT-5.5 model, reporting rapid revenue growth and strong enterprise adoption.
- 2023-01-10 product_launch OpenAI launched its new model, GPT-5.5, reporting strong initial revenue growth.
21 天有情绪数据
-
Google 发布 Gemini 3.5 Flash，用于更快的代理任务
Google 发布了 Gemini 3.5 Flash，这是一款专为速度和代理任务设计的新型 AI 模型。它被定位为 Anthropic 的 Claude Opus 4.7 和 OpenAI 的 GPT-5.5 等模型在不需要最高智能的任务上的更快、更便宜的替代品。该模型在 Google 的 Antigravity 城市建造模拟等某些应用中展示了显著的速度提升，速度提高了 12 倍，并有望用于日常 AI 工作流和复杂的、长周期的代理任务。
-
STAR-PólyaMath framework boosts AI math reasoning on benchmarks
A new multi-agent framework called STAR-PólyaMath has been introduced to improve mathematical reasoning in AI models. This system addresses issues like hallucination accumulation and memory fragmentation by employing me…
-
DexHoldem benchmark tests embodied AI in real-world Texas Hold'em
Researchers have developed DexHoldem, a new benchmark for evaluating embodied AI systems in real-world dexterous manipulation tasks, specifically playing Texas Hold'em. The system includes a ShadowHand for manipulation,…
-
Anthropic 的 Claude 在人工智能安全基准测试中领先，表现优于竞争对手
一项新的基准测试 DystopiaBench 显示，Anthropic 的 Claude 模型在安全对齐方面继续优于其他领先的 LLM。在六种反乌托邦场景中，Claude 始终拒绝生成有害内容，而 Grok 4.3、GPT-5.5、Gemini 3.1 Pro 和 DeepSeek V4 等模型在危险请求方面的合规程度各不相同。更新后的基准测试包括行为条件和合成亲密关系的新模块，并通过热力图可视化结果，显示模型在哪些方面未能通过安全测试。
-
GPT-5.5 and Claude Opus 4.7 compared for pentesting
A cybersecurity professional compared the capabilities of GPT-5.5 and Claude Opus 4.7, focusing on their practical application in pentesting rather than standard benchmarks. The user detailed their experiences using bot…
-
FutureSim benchmark tests AI forecasting with historical data
Researchers from the Max Planck Institute have introduced FutureSim, a new benchmark designed to evaluate AI agents' ability to predict real-world events using only historical web data. This method prevents agents from …
-
AI model routing slashes costs by up to 70% with smart task distribution
Developers can significantly reduce AI costs by implementing model routing, a technique that directs requests to the most cost-effective LLM capable of handling the task. This approach involves a classifier that analyze…
-
AI model restrictions questioned over governance and effectiveness
The debate around AI model restrictions highlights two key issues: governance and the effectiveness of restrictions. One perspective argues that unilateral control over powerful AI models by a single company is problema…
-
Hainan Free Trade Port reports over 2.2B yuan in zero-tariff imports
The Hainan Free Trade Port has seen over 2.2 billion yuan in goods imported under its zero-tariff policy since its implementation. Officials are working to optimize policies, align with international trade rules, and si…
-
AI agents turn bugs into exploits on new ExploitGym benchmark
A new benchmark called ExploitGym has been developed to assess AI agents' capability in transforming security vulnerabilities into actual exploits. This benchmark incorporates 898 real-world vulnerability cases across v…
-
One in seven Brits use ChatGPT for medical advice, study finds
A recent study indicates that one in seven individuals in Britain have begun using ChatGPT for medical advice, bypassing traditional general practitioners. This trend highlights a growing reliance on AI for healthcare g…
-
Frontier AI models break Capture The Flag cybersecurity competitions
The landscape of Capture The Flag (CTF) cybersecurity competitions has been fundamentally altered by the advent of advanced AI models. Initially, tools like GPT-4 offered a speed advantage, but the release of models suc…
-
AI's hottest job pays $630K; US opposes datacenters; Trump-Xi AI talks
The AI industry is experiencing a unique job market where the hottest role, with a $630K salary, is not directly involved in model development. Meanwhile, geopolitical discussions are emerging, with reports of Trump and…
-
AI agents set new records in nanoGPT training speedrun
Prime Intellect utilized advanced AI models, specifically Codex (based on GPT-5.5) and Claude Code (based on Opus 4.7), to autonomously optimize the nanoGPT training process. The AI agents conducted approximately 10,000…
-
Poetiq 的 Meta-System 在无需微调的情况下提升了 LLM 的编码性能
Poetiq 开发了一个 Meta-System，可自动创建推理 Harness，在无需任何模型微调的情况下显著提高了 LLM 在编码基准测试中的性能。该系统在 LiveCodeBench Pro 上取得了最先进的成果，将 GPT 5.5 High 的分数从 89.6% 提高到 93.9%，将 Gemini 3.1 Pro 的分数从 78.6% 提高到 90.9%。Meta-System 的 Harness 被设计为模型无关的，通过优…
-
字节跳动推出 19 款 Doubao AI 模型，定价极具竞争力
字节跳动已推出其 Doubao API，提供包括聊天、图像和视频在内的 19 款不同模型。聊天模型具有 256K 上下文窗口，并支持视觉和工具调用等高级功能。聊天模型的定价从 Seed 1.6 Flash 级别的每百万字符 0.022 美元起，极具竞争力，旗舰版 Seed 2.0 Pro 定价更高。可通过字节跳动的火山引擎Ark直接访问，该平台要求中国居民身份，或通过TokenMix等聚合器访问，后者提供更广泛的可访问性。
-
Redis creator releases DwarfStar 4 for fast local AI inference
DwarfStar 4 (DS4), a new local AI inference engine, has gained rapid popularity for its focus on integrating a single, high-performance model. Developed by Salvatore Sanfilippo, creator of Redis, DS4 is specifically opt…
-
OpenAI 的 ChatGPT 现已提供个性化财务建议
OpenAI 已为其美国 Pro 订阅用户推出了 ChatGPT 内的新个人理财功能预览。该功能允许用户通过 Plaid 安全地连接其金融账户，使 ChatGPT 能够根据用户的具体余额、交易和投资提供个性化的财务见解和指导。尽管 OpenAI 强调用户对数据的控制权以及 AI 可访问内容的限制，但关于敏感财务信息的隐私和安全问题仍然存在。
-
OpenAI releases GPT-5.5 model via API
OpenAI has announced a new model, GPT-5.5, which is now available via API. The model is designed to offer enhanced capabilities and performance for developers and users.
-
Anthropic's Opus 4.7 shows improved performance, gains 'fast mode'
Anthropic has released a faster version of its Opus 4.7 model, which some users are finding to be an improvement over previous iterations and even competing models like GPT-5.5. The enhanced performance is noted in area…