Which model leads Agentic coding (Terminal-bench 4.0)?

Q: Which model leads Agentic coding (Terminal-bench 4.0)?

Claude Opus 5.5 tops the board at 66.4

Agentic coding (Terminal-bench 4.0) Leaderboard — Model Scores

Pro Mode — raw scores, one benchmark under the lensofficial card harness

The field · 4 measured

57.466.4

4 models measured, most on the official card harness. Claude Opus 5.5 tops the board at 66.4.

4 measured·4 new this week·Lifecycle Updated Oct 16, 2026Score basis

#◆Vendor$/M in$/M out

Leader — ignites whiteField — fill cools with scoreIndependently verified only0 of 4Bars show 57.4–66.4 · ◆ solid = first-party or better · outlined = arm's-length

01Claude Opus 5.5NEWAnthropic66.4$4.0/M$20.0/M$0.181

02GPT-6 AstraNEWOpenAI58.2$10.0/M$50.0/M$0.515

03Claude Fable 5.1NEWAnthropic57.9$10.0/M$50.0/M$0.518

04Gemini 4 ArgonNEWGoogle57.4$2.0/M$10.0/M$0.105

◆ solid = independently verified · aggregator attested · vendor attributed — outlined = cross-reference · source-attributed. Dimmed ◆ = confidence inferred. “+N alts” = alternate disclosures in the drawer with why-demoted. Prices per million tokens; $/point = input price ÷ score. Every row opens its model dossier.

Verification: 4 models scored · 0 independently verified · 3 vendor cross-reference · 1 vendor-reported · 0 with source disagreement. How these tiers are assigned