AutomationBench
PulseAugur coverage of AutomationBench — every cluster mentioning AutomationBench across labs, papers, and developer communities, ranked by signal.
- used by Gemini 4 Carbon 90%
- instance of Gemini 4 Argon 90%
- competes with GPT -6 Astra 80%
- instance of GPT -6 Astra 70%
- instance of Claude (Opus 5.5 70%
- used by GPT-6.1 Sol 70%
- used by LFM2.5-24B-A2B 70%
- used by ERNIE Evolve 70%
- affiliated with Fairwind Program 70%
- competes with Claude (Opus 5.5 60%
- used by Gemini 4 Argon 60%
- used by GPT-6 Sol 60%
12 day(s) with sentiment data
-
Gemini 4 Carbon launches with promotional pricing, but task cost varies
Google has launched its Gemini 4 Carbon model with an introductory promotional price of $2 per million input tokens and $10 per million output tokens, which is half the cost of Anthropic's Claude Opus 5.5. However, thir…
-
Baidu releases Wenxin 5.5 Flash Preview with focus on speed, cost, and open-source
Baidu has launched Wenxin 5.5 Flash Preview, marking its third major announcement in a week, following the open-sourcing of Wenxin 5.5 Flash Base and the disclosure of ERNIE Evolve progress. This new model emphasizes ef…
-
Gemini 4 Carbon vs. Claude Fable 5.2: Performance and Pricing Show Mixed Results
Google's Gemini 4 Carbon and Anthropic's Claude Fable 5.2 are being compared for their performance and cost-effectiveness in coding tasks. While Gemini 4 Carbon shows strengths in specific benchmarks like Terminal-Bench…
-
Baidu releases ERNIE 5.5 Flash Preview on Qianfan, weights due Nov 25
Baidu has launched a preview of its ERNIE 5.5 Flash model, accessible via API on its Qianfan platform and also available on OpenRouter. This model boasts 280 billion parameters with 14 billion active per token and a 1 m…
-
Baidu releases Wenxin 5.5 Flash Preview with 1M context and open-source plans
Baidu has officially launched its latest foundational model, Wenxin 5.5 Flash Preview, featuring 280 billion parameters and activating 14 billion parameters per token. This new model supports a context window of up to 1…
-
Meta's Muse Ember AI model launches with low prices but questionable benchmarks
Meta Superintelligence Labs has released Muse Ember, a small multimodal model designed for agent tasks like tool calls and data extraction, with a notably low pricing structure. Despite its affordability, there are disc…
-
Google releases advanced Gemini 4 Carbon AI model for complex tasks
Google DeepMind has officially released Gemini 4 Carbon, its most advanced AI model, to the public. This model excels in complex tasks such as long-horizon agentic workflows, scientific reasoning, and cybersecurity capa…
-
Anthropic releases Claude Fable 5.2 with 30% speed boost and lower prices · 10 sources tracked
Anthropic has released Claude Fable 5.2 and Claude Mythos 5.2, featuring significant improvements in speed, cost, and performance. The new models are up to 30% faster than their predecessors and offer reduced pricing, w…
-
Liquid AI unveils efficient 1GB LFM3 model family for local agents
Liquid AI has released LFM3, a new family of open-weight models designed for efficiency and local deployment. The flagship LFM3-1.7B model, weighing approximately 1 GB, boasts a 256K-token context window and a unique tr…
-
Artificial Analysis revamps AI model rankings with new index
Artificial Analysis has updated its Intelligence Index v5, significantly altering the AI model rankings. The new index heavily relies on private, unseen test sets (60%) and equally weights Knowledge Work, Coding, Scienc…
-
SenseTime launches free multimodal agent SenseNova 6.8 Flash
SenseTime has launched its new multimodal agent model, SenseNova 6.8 Flash, which is currently available for free. This model demonstrates strong capabilities in handling complex web tasks that have challenged other lea…
-
SenseTime open-sources SenseNova Loop-9B-Preview with looped Transformer design
SenseTime has open-sourced its SenseNova Loop-9B-Preview model, which utilizes a novel Looped Transformer architecture. This design reuses middle layers with shared weights, effectively doubling their computation to inc…
-
Moonshot AI releases Kimi K3.1 Pro, deliberately downgrading cyber capabilities for safety
Moonshot AI has released Kimi K3.1 Pro, a large open-weight model with a 1-million-token context window. Notably, the company deliberately removed cyber-offense training data from this model, resulting in a measurable d…
-
Liquid AI releases LFM2.5-24B-A2B for efficient agent tasks
Liquid AI has released LFM2.5-24B-A2B, a large open-weight model designed for agentic tasks. This model utilizes a mixture-of-experts approach, activating approximately 2 billion parameters per token out of its total 24…
-
OpenAI's GPT-6.1 Sol offers near-Astra performance at lower cost on Amazon Bedrock
OpenAI has released GPT-6.1 Sol, a new model available on Amazon Bedrock, designed to offer performance comparable to GPT-6 Astra at a significantly lower cost. This model is particularly beneficial for AI agents that r…
-
Google's Gemini 4 Argon now public; pricing and benchmarks detailed
Google has made its Gemini 4 Argon model generally available, following an initial two-week period accessible only to select cyber defenders. The model is available through various Google platforms and the Gemini API, w…
-
New TURTLE benchmark evolves AI tool-use tasks for better scalability
Researchers have developed TURTLE, an Unsupervised Environment Design approach to create self-tuning benchmarks for long-horizon tool use in AI assistants. This method uses a proposer, solver, and feedback generator LLM…
-
SpecAgent protocol halves agent latency and memory use while preserving quality
Researchers have introduced SpecAgent, a new protocol for large language model agents that improves efficiency by having a smaller model draft tool-use steps and a larger model verify them. This approach reduces latency…
-
Liquid AI releases LFM2.5-24B-A2B for fast on-device agents
Liquid AI has released LFM2.5-24B-A2B, a 24 billion parameter Mixture-of-Experts (MoE) model designed for on-device agents. This model activates approximately 2 billion parameters per token and is optimized for multi-st…
-
Interlock framework enhances LLM agents for complex long-horizon tasks
Researchers have developed Interlock, a new framework designed to improve the performance of large language model agents on complex, long-horizon tasks. Interlock synthesizes these tasks into typed completion graphs, en…