OpenCodePapers

code-generation-on-terminal-bench-4-0

Code IntelligenceCode Generation
Dataset Link
Results over time
Click legend items to toggle metrics. Hover points for model names.
Leaderboard
PaperCodeAccuracyModelNameReleaseDate
Introducing Claude Sonnet 5.5✓ Link70.6Claude Sonnet 5.5 (Claude Code)2026-09-28
Introducing Claude Opus 5.5✓ Link66.36Claude Opus 5.5 (Claude Code)2026-09-22
Introducing Claude Fable 5.1 and Claude Mythos 5.160.9Claude Mythos 5.1 (max) (Claude Code --bare)2026-09-01
GPT-6 Astra: A new generation of intelligence✓ Link57.9GPT-6 Astra2026-09-03
Gemini 4 Argon: our next era of frontier intelligence✓ Link57.4Gemini 4 Argon2026-09-30
Muse Wonder: A Frontier Model For Personal & Professional AI Agents57.1Muse Wonder Full (max)2026-10-12
Every FLOP Counts: Scaling a 300B Mixture-of-Experts LING LLM without Premium GPUs✓ Link46.03Ling-3.1-flash (LingCode)2025-03-07
DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression✓ Link42.3DeepSeek-V4.1-Flash-1017 (Claude Code)2026-09-17
Introducing Claude Haiku 5.539.2Claude Haiku 5.5 (max effort) (Claude Code)2026-10-07
Introducing Grok 4.737.6Grok 4.7 (xhigh)2026-09-21
MiMo-V2.6: Scaling Reinforcement Learning Towards Self-Improvement✓ Link34.9MiMo-V2.6-Pro-RL2026-10-08
Step 5 Preview33.3Step 5 Preview (High)2026-09-20
DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression✓ Link31.2DeepSeek-V4.1-Flash Max (DeepSeek Harness (Minimal))2026-09-17
Mistral Large 428.3Mistral Large 4 Preview2026-10-06
Introducing SWE-2: Pushing the Pareto Frontier27.3SWE-22026-09-10
Terminal-RSI: Recursive Self-Improving Terminal Agents via Recursive Self-Rewrite and Policy-Refine19.84Terminal-RSI-27B2026-10-15
Introducing Gemini 3.8 Flash and 3.8 Flash Cyber✓ Link19.1Gemini 3.8 Flash2026-09-02
Scaling Trajectories for Complex Tasks through Recursive Self-Rewrite6.1Recursive Self-Rewrite (Qwen-3.8-27B) (general harness)2026-10-02
Self-Supervised Scaling of Terminal Environments for Scientific Domains3.54Qwen3.8-27B (SWR SFT-3000)2026-10-02