GPT-6.1 Sol

Google's Gemini 4 Carbon has become the first non-OpenAI model in a while to top the MathArena math benchmark, displacing GPT-6.1 Sol, with Anthropic's Claude Fable 5.2 taking second place. However, its evaluation cost runs roughly 15 times that of GPT-6.1 Sol, and its most common failure mode is exceeding the time limit while almost never abandoning a task. The best open model, Moonshot AI's Kimi K3.1 Pro, trails at 54.3%.

Deepnote released the Agent Initiative Benchmark, which drops proactive agents into a simulated company Slack and scores whether they know when to intervene, what to say, and when to stay silent. GPT-6 Astra led at 8.97 out of 10, with GPT-6.1 Sol close behind at 8.65 at just $0.11 per task, versus $1.87 per task for third-placed Claude Fable 5.2. The open-source dataset and benchmark code give builders a reproducible basis for choosing models for proactive agent use cases.

Amplitude disclosed that its proactive product agent Wave is roughly 80% powered by OpenAI models, including GPT-6.1 Sol, GPT-6 Luna, and GPT-Image-2.5 Flare, converting product insights into implementation plans that can be handed to coding agents such as Codex. The company claims this compresses product improvement cycles from about a month to 1-2 days, with personalized suggestion cards on its own homepage driving a 160% lift in card-driven sends.

Amplitude disclosed that its proactive product agent Wave is roughly 80% powered by OpenAI models, including GPT-6.1 Sol, GPT-6 Luna and GPT-Image-2.5 Flare, converting product insights into implementation plans handed to coding agents such as Codex. The company says improvement cycles shrink from about a month to 1-2 days, with a 160% lift in card-driven sends on its own homepage.

Google DeepMind has shipped Gemini 4 Hydrogen, its third new model in six weeks, introducing an xhigh reasoning tier above high. It scores 54 on the Artificial Analysis Intelligence Index, just one point behind Claude Sonnet 5.5 and GPT-6.1 Sol, at roughly 32% of Sonnet's cost but about 2.5x that of GPT-6.1 Sol, with standout visual reasoning and non-hallucination performance.

Cogent Security has launched enterprise Cyber Defense Factories that use AI agents to autonomously close the loop from vulnerability discovery to verified remediation. The agents combine GPT‑6.1 Sol and GPT‑5.6 with enterprise context and controls to assess exposures, execute permitted changes, and confirm risks are eliminated. Cogent is also developing autonomous discovery and validation toward a continuously learning closed-loop defense system.

AI/ML API pitted Claude Fable 5.2 against GPT-6.1 Sol to rebuild the Cars Piston Cup finale in 3D voxels, with Fable 5.2 spending $108.87 versus $13.02 for Sol. The cost gap reflects Fable's strict voxel adherence, modeling even smoke as geometry, while Sol rendered the crash car as a dark silhouette. The side-by-side highlights trade-offs in style compliance and visual detail between the two models.