Ledger on GitHub. Hub here. Branches elsewhere.
drop-2026-09-10-kr-close
| claim | label | order | numbers | URL |
|---|---|---|---|---|
| Claude Opus 5 is the premium coding leader, with $5 / $25 pricing and top August 2026 coding/agent leaderboard placement. | 🟢 robust | 1 | $5 / $25; 58/63; 63 intelligence | https://rubyonrails.org/2026/8/17/agents-on-rails-grok-4-6-glm-5-3-gemini-3-7-flash-and-opus-4-8 https://lumichats.com/blog/best-ai-model-right-now-august-2026-ranked |
| GPT-5.6 Sol is top-tier on reasoning and agents, but trails Claude Opus 5 on the August 2026 Intelligence Index. | ⚠️ sensitive | 2 | 61; 57.78 agentic; 88.01% TB2.1 | https://lumichats.com/blog/best-ai-model-right-now-august-2026-ranked https://rohitai.com/blog/best-ai-models-2026-openai-anthropic-google-xai-deepseek |
| Grok 4.6 is a cheaper frontier coding/agent option, with strong Terminal-Bench and competitive campaign results. | 🟢 robust | 3 | $2 / $6; 88.39% TB2.1; 52/63 | https://rohitai.com/blog/best-ai-models-2026-openai-anthropic-google-xai-deepseek https://rubyonrails.org/2026/8/17/agents-on-rails-grok-4-6-glm-5-3-gemini-3-7-flash-and-opus-4-8 |
| Gemini 3.7 Flash is the speed/latency leader in the August 2026 snapshot, not the overall quality leader. | 🟢 robust | 4 | 340.07 tok/s; 9.83 s; 45/63 | https://rohitai.com/blog/best-ai-models-2026-openai-anthropic-google-xai-deepseek https://rubyonrails.org/2026/8/17/agents-on-rails-grok-4-6-glm-5-3-gemini-3-7-flash-and-opus-4-8 |
| Qwen3.8-Flash-Next launched as a low-cost, long-context API model with pricing around $0.16 input and $0.47 output per 1M tokens. | 🟢 robust | 5 | $0.16 / $0.47; ~1M context | https://www.datacamp.com/blog/qwen3-8-flash-next https://www.requesty.ai/blog/open-weight-frontier-august-2026-glm-qwen-hy4 |
| Qwen3.8-Flash-Next also shows low-latency API listings, but that latency figure is only single-source here. | ⚠️ sensitive | 6 | TTFT 2.67 s | https://artificialanalysis.ai/models/qwen3-8-flash-next |
| GLM-5.3 Flash is the very low-cost open-weight/API option, with pricing around $0.075–$0.08 input and $0.25 output per 1M tokens. | ⚠️ sensitive | 7 | $0.075–$0.08 / $0.25; 1M context | https://openrouter.ai/z-ai/glm-5.3-flash https://developer.puter.com/ai/z-ai/glm-5.3-flash/ |
| GLM-5.3 Flash is also described as open-weight with 1M context and MIT licensing. | 🟢 robust | 8 | 320B total; 18B active; 1M context | https://enterprisedna.co/resources/news/z-ai-ox-alpha-glm-53-flash-open-weight-enterprise-2026/ |
| August 2026 saw multiple open-weight launches in a short window, including GLM-5.3-Flash, Qwen3.8-Flash, Hy4 Preview, MiniMax M3, and DeepSeek V4-Flash-Vision-Exp. | 🟢 robust | 9 | 5 releases; 9 days | https://www.requesty.ai/blog/open-weight-frontier-august-2026-glm-qwen-hy4 |
| DeepSeek V4-Flash-Vision-Exp is described as a stronger multimodal agent model, but the claim is promotional and low-confidence. | ⚠️ sensitive | 10 | qualitative only | https://www.requesty.ai/blog/open-weight-frontier-august-2026-glm-qwen-hy4 |