isitsota / categories / Coding › Agentic
Coding · Oct 14, 2026
Who is SOTA in Coding › Agentic?
Claude Opus 5.5 & Claude Sonnet 5.5 & GLM 5.3 & GLM 5.3 Flash & GPT-6 Luna
Agentic terminal work on Terminal-Bench 4.0. Version 2.0 stays in the record and is not mixed into the 4.0 score. As of Oct 14, 2026, Claude Opus 5.5 & Claude Sonnet 5.5 & GLM 5.3 & GLM 5.3 Flash & GPT-6 Luna holds Coding › Agentic. On the cost-quality frontier for…
Standings · pareto
The board
| # | Model | State | Confidence |
|---|---|---|---|
| 1 | AClaude Opus 5.5 Anthropic | Pareto | medium |
| 2 | AClaude Sonnet 5.5 Anthropic | Pareto | medium |
| 3 | OGPT-6.1 Sol OpenAI | Near-SOTA | medium |
| 3 | OGPT-6 Astra OpenAI | Near-SOTA | medium |
| 5 | AClaude Fable 5.1 Anthropic | Not SOTA · #5 | medium |
| 6 | AClaude Opus 5 Anthropic | Not SOTA · #6 | medium |
| 7 | OGPT-6 Sol OpenAI | Not SOTA · #7 | medium |
| 8 | AClaude Fable 5 Anthropic | Not SOTA · #8 | medium |
| 9 | ZGLM 5.3 Zhipu AI | Pareto | medium |
| 10 | xGrok 4.7 xAI | Not SOTA · #10 | medium |
| 11 | OGPT-5.6 Sol OpenAI | Not SOTA · #11 | medium |
| 12 | ZGLM 5.3 Flash Zhipu AI | Pareto | medium |
| 13 | QQwen3 8 Max 0902 Alibaba / Qwen | Not SOTA · #13 | medium |
| 14 | AClaude Opus 4.8 Anthropic | Not SOTA · #14 | medium |
| 15 | OGPT-5.6 Terra OpenAI | Not SOTA · #15 | medium |
Pareto
Quality against blended price
Evidence recipe
What can move this crown
- Terminal-Bench 4.0 — primary, weight 1
- Artificial Analysis Agentic Index — secondary, weight 0.4
- Terminal-Bench 2.0 — display, weight 0