TeamBench - Team without Planner: 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 Planner, an Executor that sees only the brief works with a Verifier that holds the specification; 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 an Executor and a Verifier of the same model, the Verifier holding the specification and the Executor seeing only the brief; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Claude Opus 4.735.6#85
2GPT-5.4 Mini25.6#406
3Gemma 4 31B24.4#382
4GPT-5.423.3#135
5Claude Haiku 4.518.9#583
6Gemini 3.1 Pro (Preview)16.7#99
7Gemini 3 Flash14.4#175
8GPT-OSS-20B12.2#1183
9Claude Sonnet 4.610#173
10Gemini 3.1 Flash Lite8.9#371

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-planner · How It Works · Data refreshed daily, snapshot 2026-11-21.