Benchmark Almanac

Harbor-Index

GPT-5.5 28.1% leads since Jun 2026 In use

Agents solve about 80 hard tasks drawn from 29 benchmarks run through Harbor adapters, such as HLE, GSO and SWE-bench Pro. Each task keeps its source verifier.

Version 1.0 (June 2026) had 82 tasks. Releases 1.1 to 1.4 in July 2026 fixed and re-pinned tasks to 80, each with its own frozen leaderboard. No pair passed 30% at launch.

Made by Harbor and Laude Institute (Terminal-Bench team) Released Jun 2026 Terminal tasks All tasks public Last in a launch post Aug 2026 Last model added Aug 2026 Scores refreshed 2 Nov 2026
Level
Research frontier Researchers, on open problems or tasks that take days
Scored
Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
Answers
Agent task Many steps in a terminal, browser, desktop, app, game or robot.
harbor-framework/harbor-index 44 stars Dataset Leaderboard tbench.ai
9
Models scored on its own leaderboard
1
Launch post lists it, from 1 lab

GPT-5.5 leads with 28.1% among models released by Jun 2026

Best score per model from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.

The newest model with a score here came out in Jun 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 28.1 GPT-5.5
  2. 20.7 Claude Opus 4.8
  3. 13.4 Gemini 3.1 Pro
  4. 9.80 GLM-5.2
  5. 8.50 Kimi K2.6
  6. 6.10 MiniMax-M3
  7. 4.90 DeepSeek-V4-Pro
  8. 4.90 Qwen3.7-Max
  9. 2.40 MiMo-V2.5-Pro
  1. 1 GPT-5.5 28.1%
  2. 2 Claude Opus 4.8 20.7%
  3. 3 Gemini 3.1 Pro 13.4%
  4. 4 GLM-5.2 9.80%
  5. 5 Kimi K2.6 8.50%
  6. 6 MiniMax-M3 6.10%
  7. 7 DeepSeek-V4-Pro 4.90%
  8. 8 Qwen3.7-Max 4.90%
  9. 9 MiMo-V2.5-Pro 2.40%
#ModelScoreSettingDateSource
1 GPT-5.5 28.1% Codex CLI Apr 2026 Official leaderboard
2 Claude Opus 4.8 20.7% Claude Code May 2026 Official leaderboard
3 Gemini 3.1 Pro 13.4% Gemini CLI Feb 2026 Official leaderboard
4 GLM-5.2 9.80% Terminus 2 Jun 2026 Official leaderboard
5 Kimi K2.6 8.50% Terminus 2 Apr 2026 Official leaderboard
6 MiniMax-M3 6.10% Terminus 2 Jun 2026 Official leaderboard
7 DeepSeek-V4-Pro 4.90% Claude Code Apr 2026 Official leaderboard
8 Qwen3.7-Max 4.90% Claude Code May 2026 Official leaderboard
9 MiMo-V2.5-Pro 2.40% Terminus 2 Apr 2026 Official leaderboard

Labs report scores for 1 model in their cards. We plot every source in the charts below.

The best score, 28.1%, costs $2.17 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Google DeepMind Moonshot AI OpenAI Xiaomi Corp MiniMax Z.ai Cost frontier One model at each effort level
0% 10% 20% 30% $0.01 $0.10 $1.00 $10 Cost per task, log scale GPT-5.5 GPT-5.5 Gemini 3.1 Pro GLM-5.2 Kimi K2.6 MiniMax-M3 MiMo-V2.5-Pro 0% 10% 20% 30% $0.01 $0.10 $1.00 $10 Cost per task, log scale GPT-5.5 GPT-5.5
ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.5 Codex CLI 28.1% $2.17
Claude Opus 4.8 Claude Code 20.7% $3.28
GPT-5.5 Terminus 2 19.7% $1.89
Claude Opus 4.8 Terminus 2 15.8% $3.57
Gemini 3.1 Pro Gemini CLI 13.4% $0.90
Gemini 3.1 Pro Terminus 2 10.7% $1.08
GLM-5.2 Terminus 2 9.80% $0.63
Kimi K2.6 Terminus 2 8.50% $0.40
GLM-5.2 Claude Code 8.50% $2.50
Kimi K2.6 Claude Code 6.10% $2.33
MiniMax-M3 Terminus 2 6.10% $0.22
DeepSeek-V4-Pro Claude Code 4.90% $2.16
Qwen3.7-Max Claude Code 4.90% $2.45
Qwen3.7-Max Terminus 2 4.90% $0.44
DeepSeek-V4-Pro Terminus 2 3.70% $0.43
MiniMax-M3 Claude Code 3.70% $0.80
MiMo-V2.5-Pro Terminus 2 2.40% $0.05
MiMo-V2.5-Pro Claude Code 2.40% $0.60

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Own leaderboard Not tested since Jun 2026
GPT-5.5 28.1%

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Tencent Hy4 preview GPT-5.5 Gemini 3.1 Pro 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Tencent Hy4 preview GPT-5.5

One lab put it in a launch post in the last six months

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

Tencent
DateLaunch postScoreAs the post writes it
28 Aug 2026 Tencent Hy4 preview 39.6 Harbor-Index

More in Terminal agents

All 18 in Terminal agents
  • Terminal-Bench 4.0 In use Agents complete 66 hard tasks in containerized terminals across software, ML, science, security, and operations. Test scripts verify each result, scored as resolution rate. Claude Opus 5.5 64.8%
  • SkillsBench In use 86-87 containerized expert tasks in 11 domains, run with and without Agent Skills. Verifier tests give pass rate and skill lift. GPT-5.5 67.3%
  • Terminal-Bench Hard In use Agents complete 44 hard tasks from Terminal-Bench Core in the Terminus 2 harness. Artificial Analysis scores pass@1 averaged over 3 repeats. Gemma 4 31B IT 36%
  • CompileBench Not in model cards Agents build real open-source projects through dependency problems, legacy toolchains and complex build systems, for example compiling 22-year-old code. Claude Opus 4.1 100%
  • Long-Horizon-Terminal-Bench 2.0 Not in model cards Agents work for up to 3 hours per task on 77 container tasks, including 30 tasks that rebuild a scientific tool from its outputs. Hidden verifiers give partial-credit rewards. GPT-6 Astra 69.2%
  • ITBench Not in model cards Agents handle IT automation tasks in Kubernetes environments: site reliability incidents, compliance and security checks, and cloud cost work. Gemini 3 Pro 58.3%