Independent review lifts coding-agent accuracy at modest extra cost
A critic review gains 5.9–9.5 points on Terminal-Bench and 3.9–6.6 on SWE-bench Verified at about baseline cost.
48
Product
Terminal-Bench
- Mentions · 7d0
- Mentions · 30d1
- All mentions2
- First seenSep 9, 2026
- Last mentioned8d ago
Timeline
2 articles · newest firstT1: Terminal Agent Reinforcement Learning for Long-Horizon Tasks
T1, a 122B MoE terminal agent trained with reinforcement learning, reaches 64.0% on Terminal-Bench 2.1, surpassing GPT-5.4 and GLM-5.1 on long-horizon tasks.
55
Appears with
- SWE-bench Verifiedproduct · 1
- VisualBenchproduct · 1
- GPT-5.4ai model · 1
- GLM-5.1ai model · 1
- T1ai model · 1
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.