Benchmark
TBench
Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages.
- Models measured
- 89
- Scores recorded
- 234
- Sources
- 1
- IQ weight
- 0.93
Score distribution
- 42.9%
- Mean
- 42.2%
- Median
- 22.9%
- Std dev
- 14.2%
- 10th percentile
- 75.1%
- 90th percentile
- 0%
- Min
This benchmark has a weight of 0.93 for chat models, measured over 89 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | |
|---|---|---|
| 1 | Inkling Smallterminal-bench-4.0, mini-swe-agent· Terminal-Bench | 91% |
| 2 | Inkling Smallterminal-bench-4.1, mini-swe-agent· Terminal-Bench | 91% |
| 3 | GPT-6 Astraterminal-bench-2.1, Codex· Terminal-Bench | 87.4% |
| 4 | GPT 5.5terminal-bench-2.0, NexAU-AHE· Terminal-Bench | 84.7% |
| 5 | GPT 5.5terminal-bench-2.1, Codex· Terminal-Bench | 83.2% |
| 6 | GPT 5.5terminal-bench-2.0, Capy· Terminal-Bench | 83.1% |
| 7 | GPT 5.5terminal-bench-2.0, Codex CLI· Terminal-Bench | 82.2% |
| 8 | Claude Opus 4.7terminal-bench-2.0, WOZCODE· Terminal-Bench | 80.2% |
| 9 | Gemini 3.1 Pro Previewterminal-bench-2.0, TongAgents· Terminal-Bench | 80.2% |
| 10 | Grok 4.5terminal-bench-2.1, Cursor CLI· Terminal-Bench | 79.3% |
| 11 | Claude Opus 4.8terminal-bench-2.1, Claude Code· Terminal-Bench | 78.9% |
| 12 | GPT-5.6 Terraterminal-bench-2.1, Codex· Terminal-Bench | 78.4% |
| 13 | GPT-5.3 Codexterminal-bench-2.0, SageAgent· Terminal-Bench | 78.4% |
| 14 | GPT 5.5terminal-bench-2.1, Terminus 2· Terminal-Bench | 78% |
| 15 | GPT-5.3 Codexterminal-bench-2.0, Droid· Terminal-Bench | 77.3% |
| 16 | Claude Opus 4.6terminal-bench-2.0, Meta-Harness· Terminal-Bench | 76.4% |
| 17 | Muse Spark 1.1terminal-bench-2.1, mini-SWE-agent· Terminal-Bench | 76.2% |
| 18 | GPT-5.3 Codexterminal-bench-2.0, CodeBrain-1.5· Terminal-Bench | 75.8% |
| 19 | GPT-5.3 Codexterminal-bench-2.0, Codelia· Terminal-Bench | 75.7% |
| 20 | GPT-5.6 Lunaterminal-bench-2.1, Codex· Terminal-Bench | 75.7% |
| 21 | Claude Opus 4.6terminal-bench-2.0, Capy· Terminal-Bench | 75.3% |
| 22 | GPT-5.3 Codexterminal-bench-2.0, Simple Codex· Terminal-Bench | 75.1% |
| 23 | Gemini 3.1 Pro Previewterminal-bench-2.0, Terminus-KIRA· Terminal-Bench | 74.8% |
| 24 | Claude Opus 4.6terminal-bench-2.0, Terminus-KIRA· Terminal-Bench | 74.7% |
| 25 | GPT-5.3 Codexterminal-bench-2.0, Mux· Terminal-Bench | 74.6% |
| 26 | Gemini 3 Pro Previewterminal-bench-2.1, Terminus 2· Terminal-Bench | 73.9% |
| 27 | Claude Opus 4.6terminal-bench-2.0, MAYA-V2· Terminal-Bench | 72.1% |
| 28 | Claude Opus 4.6terminal-bench-2.0, TongAgents· Terminal-Bench | 71.9% |
| 29 | GPT-5.3 Codexterminal-bench-2.0, spoox-o-m· Terminal-Bench | 71.5% |
| 30 | Claude Opus 4.6terminal-bench-2.0, Droid· Terminal-Bench | 69.9% |
| 31 | Gemini 3 Pro Previewterminal-bench-2.0, Ante· Terminal-Bench | 69.4% |
| 32 | GPT-5.3 Codexterminal-bench-2.0, IndusAGI Coding Agent· Terminal-Bench | 69.1% |
| 33 | Claude Opus 4.7terminal-bench-2.1, Claude Code· Terminal-Bench | 68.9% |
| 34 | Claude Opus 4.6terminal-bench-2.0, Crux· Terminal-Bench | 66.9% |
| 35 | Claude Opus 4.6terminal-bench-2.0, Mux· Terminal-Bench | 66.5% |
| 36 | GPT-5.2 Codexterminal-bench-2.0, Deep Agents· Terminal-Bench | 66.5% |
| 37 | Claude Opus 4.7terminal-bench-2.1, Terminus 2· Terminal-Bench | 66.1% |
| 38 | GPT 5.5terminal-bench-2.0, clnkr· Terminal-Bench | 66.1% |
| 39 | Gemini 3 Pro Previewterminal-bench-2.1, Gemini CLI· Terminal-Bench | 65.8% |
| 40 | Gemini 3.1 Pro Previewterminal-bench-2.1, Gemini CLI· Terminal-Bench | 65.8% |
| 41 | Claude Opus 5.5terminal-bench-4.1, Claude Code· Terminal-Bench | 65.8% |
| 42 | Gemini 3.1 Pro Previewterminal-bench-2.1, Terminus 2· Terminal-Bench | 65.6% |
| 43 | Gemini 3 Pro Previewterminal-bench-2.0, SageAgent· Terminal-Bench | 65.2% |
| 44 | GPT 5.2terminal-bench-2.0, Droid· Terminal-Bench | 64.9% |
| 45 | Claude Opus 5.5terminal-bench-4.0, Claude Code· Terminal-Bench | 64.9% |
| 46 | GPT-5.3 Codexterminal-bench-2.0, Terminus 2· Terminal-Bench | 64.7% |
| 47 | Gemini 4 Carbonterminal-bench-4.1, Antigravity CLI· Terminal-Bench | 64.6% |
| 48 | Gemini 3 Flashterminal-bench-2.0, Junie CLI· Terminal-Bench | 64.3% |
| 49 | Claude Opus 4.5terminal-bench-2.0, Droid· Terminal-Bench | 63.1% |
| 50 | Claude Opus 4.6terminal-bench-2.0, Terminus 2· Terminal-Bench | 62.9% |
Showing 1–50 of 234 results.
Tokens the model generated per task, as measured by the source. Comparable within this benchmark only.
Data credits: Terminal-Bench (Laude Institute), Apache-2.0 (licence)