TeamBench - Team without Verifier: leaderboard

TeamBench (2026) tests agent coordination under operating-system-enforced role separation on a 90-task stratified subset graded by deterministic scripts, every role filled by the same model at temperature 0. Without the Verifier, a Planner that reads the specification guides an Executor that edits the workspace; pass rate in percent, higher is better.

Metric: Pass rate (%) on the 90-task stratified TeamBench leaderboard subset (operating-system-enforced role separation in Docker sandboxes; a run passes when every check of the task's deterministic shell-script grader passes, and a run failing only the attestation check counts as a pass), temperature 0, with a Planner that reads the specification and an Executor that edits the workspace, with no Verifier attestation; higher is better. Source: arxiv.org. Saturation forecast: Around March 2027. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GPT-5.434.4#135
2Claude Opus 4.733.3#85
3Gemini 3 Flash27.8#175
4Gemini 3.1 Pro (Preview)25.6#99
5GPT-5.4 Mini24.4#406
6Gemma 4 31B20#382
7Gemini 3.1 Flash Lite17.8#371
8GPT-OSS-20B7.8#1183
9Claude Sonnet 4.66.7#173
10Claude Haiku 4.51.1#583

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=teambench-team-without-verifier · How It Works · Data refreshed daily, snapshot 2026-11-21.