Kimi K3 excels on new Frontier Code v1.1 bench but flops on Terminal Bench v4
Ok_Warning2146 · reddit · 2026-10-20
A Reddit discussion compares two agentic coding benchmarks: the new Frontier Code v1.1, which claims to measure agentic software engineering, and Terminal Bench v4.
The spark: Kimi K3 scores very well on Frontier Code v1.1 while performing poorly on Terminal Bench v4, raising questions about what different capabilities the two benchmarks actually measure.
More from Models
- What Gen Models Run Well on a 3060/4060? Reddit Weighs Local Setups — Haghiri75 · 2026-10-20
- Sol 6.1 caught cheerfully talking itself through task assignment in agent logs — TheZachMueller · 2026-10-20
- Unreleased Moonshot model Kimi-K3.1 spotted in sglang PR, specs unconfirmed — TheZachMueller · 2026-10-20
- Claude Opus 5.5 tops TxBench antibody-discovery benchmark at 67.7%, GPT-6 Astra leads oligo tasks — kenbwork · 2026-10-20
- stealthbench adds 10 EDR-guarded tasks: models' opsec behavior is abysmal — dyn___ · 2026-10-20
- "Everyone has a little Terry Tao in their pocket now" — Reza_Zadeh · 2026-10-20