# Rings AI Models and Published Evaluations Rings AI is by monday.com. Stay current with leading frontier models from top AI labs. Explore Rings AI’s model lineup, compare published evaluations, and choose the intelligence for your agents. We keep Rings AI current with leading frontier models from top AI labs, adding new releases as they become available and ready for your agents. ## Published Evaluations Provider comparisons, not Rings AI tests. Higher is better within each chart; different benchmarks and settings measure different things. ### Multi-Step Coding: Terminal-Bench 4.0 Tasks completed. - Claude Opus 5.5 (xhigh): 66.4% - GPT-6 Astra (high): 57.9% - Claude Fable 5.1: 55.8% - Claude Opus 5: 52.3% - GPT-5.6 Sol: 37.3% September 22 comparison. OpenAI results are reproduced by Anthropic. Standard error: ±2.6 points for Opus 5.5; ±1.6–2 for other Claude models. Production safeguards can route some tasks to fallback models. [Anthropic · Opus 5.5 release](https://www.anthropic.com/claude-opus-5-5) ### Code Ready to Merge: FrontierCode 1.1 · Main Accuracy. - Claude Opus 5.5: 54.4% - Claude Sonnet 5.5 (xhigh): 52.1% - GPT-6 Sol: 49.3% - Claude Sonnet 5: 42.4% Sonnet 5.5 scores 52.1% at xhigh and 46.2% at max; this chart shows xhigh. These are the release report’s comparison settings, not identical effort settings across models. [Anthropic · Sonnet 5.5 release](https://www.anthropic.com/claude-sonnet-5-5) ### Everyday Knowledge Work: GDPval-AA v2.1 Score. - Claude Sonnet 5.5: 1840 - Claude Haiku 5.5: 1620 - GPT-6 Luna: 1437 - Claude Haiku 4.5: 735 Professional tasks across 44 occupations. Scores come from the Haiku 5.5 comparison, including its Sonnet reference result. These scores are not percentages. [Anthropic · Haiku 5.5 release](https://www.anthropic.com/claude-haiku-5-5) ### Work on a Computer: OSWorld 2.1 · Offline Subset Task performance. - Claude Sonnet 5.5: 83.9% - Claude Haiku 5.5: 72.4% - GPT-6 Luna: 48.9% - Claude Haiku 4.5: 15.7% All four values use the offline subset reported in the Haiku 5.5 release. Do not compare these directly with full-set or other OSWorld versions. [Anthropic · Haiku 5.5 release](https://www.anthropic.com/claude-haiku-5-5) ### Work across Apps: AutomationBench 1.0.6 Score. - GPT-6 Sol (xhigh): 33.2% - Claude Fable 5.1 (max · Opus 5 fallback): 31.4% - GPT-6 Astra (low): 30.3% - Claude Opus 5 (max): 26.9% Workflow comparison across 47 tools, reproduced in OpenAI’s release. Effort differs by row; Astra is shown at low, not its highest setting. Fable uses Opus 5 fallbacks. [OpenAI · GPT-6 Sol and Luna release](https://openai.com/index/introducing-gpt-6-sol-and-luna/) ## Model Catalog ### GPT-6 Astra OpenAI. A frontier option for demanding reasoning and agentic work. [Docs](https://developers.openai.com/api/docs/models/gpt-6-astra) ### GPT-6.1 Sol OpenAI. For complex coding, computer use, and professional work. [Docs](https://developers.openai.com/api/docs/models/gpt-6.1-sol) ### GPT-5.6 Terra OpenAI. A balance of intelligence and cost for everyday agent work. [Docs](https://developers.openai.com/api/docs/models/gpt-5.6-terra) ### GPT-6 Luna OpenAI. For fast, focused work and frequent agent tasks. [Docs](https://developers.openai.com/api/docs/models/gpt-6-luna) ### Claude Opus 5.5 Anthropic. For complex coding, research, and work that takes multiple steps. [Docs](https://www.anthropic.com/claude-opus-5-5) ### Claude Fable 5.1 Anthropic. A capable option for challenging reasoning and agent tasks. [Docs](https://www.anthropic.com/claude-fable-and-mythos-5-1) ### Claude Opus 5 Anthropic. For in-depth analysis and complex agent workflows. [Docs](https://www.anthropic.com/news/claude-opus-5) ### Claude Sonnet 5.5 Anthropic. For everyday coding, research, and creating documents, slides, and spreadsheets. [Docs](https://www.anthropic.com/claude-sonnet-5-5) ### Claude Haiku 5.5 Anthropic. For quick summaries, repetitive work, and focused background tasks. [Docs](https://www.anthropic.com/claude-haiku-5-5) ### NVIDIA Nemotron 3 Ultra NVIDIA. NVIDIA’s open-weight option for reasoning and agentic tasks. [Docs](https://research.nvidia.com/labs/nemotron/Nemotron-3-Ultra/) ### NVIDIA Nemotron Lightning 3.5 NVIDIA. A lighter Nemotron option for responsive everyday work. [Docs](https://catalog.ngc.nvidia.com/orgs/nim/nvidia/models/nemotron-3.5-lightning/hf-3db7814) ### Kimi K3 Moonshot AI. For work that combines broad context with reasoning and tools. [Docs](https://huggingface.co/moonshotai/Kimi-K3) ### Kimi K2.6 Moonshot AI. An open-weight option for coding and tool-assisted tasks. [Docs](https://huggingface.co/moonshotai/Kimi-K2.6) ### DeepSeek V4 Pro DeepSeek. For demanding reasoning and coding tasks. [Docs](https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro) ### DeepSeek V4 Flash DeepSeek. A smaller V4 option for frequent agent tasks. [Docs](https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash) ### Grok 4.1 Fast (Non-Reasoning) xAI. For fast responses without a separate reasoning phase. [Docs](https://x.ai/news/grok-4-1-fast) ### Claude Opus 4.8 Anthropic. An earlier Opus option for reasoning and agent work. [Docs](https://www.anthropic.com/news/claude-opus-4-8) ### Claude Haiku 4.5 Anthropic. A compact Claude option for responsive everyday tasks. [Docs](https://www.anthropic.com/news/claude-haiku-4-5) ## Will Rings AI Keep Adding New Models? We keep Rings AI current with leading frontier models from top AI labs, adding new releases as they become available and ready for your agents. We continue to expand support across model providers. Timing depends on provider availability and compatibility checks; the app’s model selector shows what you can use today. ## How Should I Choose a Model? Start with a few real tasks. Compare answer quality, tool use, time to completion, and usage. A benchmark can narrow your shortlist, but your team’s work is the most useful test. ## Are These Rings AI Benchmark Results? No. The scores are published by model providers. They use specific tools, effort settings, and evaluation environments. They are reference points, not a prediction of an agent’s success rate in Rings AI. ## What Does Rings AI Super Intelligence Add to the Model? The model supplies intelligence. Rings AI Super Intelligence brings context, tools, memory, goals, and review into the work around it. Choose a capable model, then give your agent the context and boundaries its task needs. [Explore the model catalog](https://staging.getrings.ai/models/). [Explore Rings AI Super Intelligence](https://staging.getrings.ai/agent-harness/).