codex-cli + gpt-5.6-terra
codex-cli harness · ChatGPT Plus ($20/mo)
Clears tier 3 of 5 (partial at tiers 4 and 5) on Real-world issues
Capability mean across 2 graded suite families.
What the tiers mean
"Clears" marks the highest rung where every task at it and below was solved. Tier 1: localized single-file bug · tier 3: cause in a different module than the symptom · tier 5: problems that took a human hours. A cleared rung solved every task on it; a partial rung solved some. The headline stops at the last unbroken rung, so a solved tier above a half-solved one counts only as partial.
What the levels mean
Level 1: a full brief naming the surface and the acceptance · level 3: the report as filed · level 5: a vague vibe report in a non-technical user's words. Context falls as the level rises, so vaguer is harder. A cleared level fixed every bug at it and below; a partial level fixed some. The headline stops at the last unbroken level, so a fix above a half-cleared level counts only as partial.
7 of this suite's tasks were credited rather than run: on a suite that asks the same problem at several levels of detail, solving it from the vaguest description credits the more detailed ones instead of asking again.
Per-suite results
11 of 15 tasks solved - grouped by tier, easiest first; hover a square for its time and turns.
Run on codex-cli codex-cli 0.147.0, model gpt-5.6-terra (-c model_reasoning_effort=medium), week 2026-W33.
One run of this suite consumes 2.1% of ChatGPT Plus.
How the plan cost was measured
Measured per suite run (estimated-from-telemetry); an absolute cost on this plan, not comparable with another agent's plan. Measured in a declared quiet window on a shared subscription: operator ChatGPT Plus login shared between eb-runner and the claude host; graded runs only inside the scheduled quiet window, all recurring tasks blocked for its duration.
17 of 25 tasks solved (7 credited without running) - each row is one bug, vaguest report first and context growing to the right; hover a square for its time and turns.
Run on codex-cli codex-cli 0.147.0, model gpt-5.6-terra (-c model_reasoning_effort=medium), week 2026-W33.
One good solve on this suite consumes 0.2% of ChatGPT Plus on average.
How the plan cost was measured
Counted over each issue's first solve (5 solves); the walk's failed harder cells and confirmation runs are the benchmark's own search cost and are excluded (full walk: 3.0% over 18 cells). Measured per cell run (estimated-from-telemetry); an absolute cost on this plan, not comparable with another agent's plan. Measured in a declared quiet window on a shared subscription: operator ChatGPT Plus login shared between eb-runner and the claude host; graded runs only inside the scheduled quiet window, all recurring tasks blocked for its duration.
Task outputs from these runs are withheld: publishing them would reveal the private suite. Per-task scores, times and turns are in the chips above.
Trajectory
One point per grading week - a re-run within a week shows only its latest execution. Grades compare only within a suite version, so each version gets its own line; while a suite is being retired both versions run, and the superseded one is dashed. The filled point is the run the scorecard shows today. Click a point or week for plan usage. All agents over time →
Weekly heads
Newest week first. Click a row for usage.
| Week | Grade | Capability | Usage |
|---|---|---|---|
| 2026-W33 · v3now | D | 53% | 2.1% |
| 2026-W32 · v3 | F | 28% | 1.4% |
Frequently asked questions
Is gpt-5.6-terra worth it on ChatGPT Plus?
On our private real-world suite (real-world@v3), gpt-5.6-terra earned grade D with 53% capability, clearing every problem up to tier 3 of 5 - a problem whose cause sits in a different module than the symptom and landing some tier 5 problems. It runs on ChatGPT Plus at $20/month (public price as of 2026-07-20). One full suite run consumed 2.1% of the plan's week window. Measured 2026-08-10.
How good is gpt-5.6-terra at real-world coding?
gpt-5.6-terra clears tier 3 of our five-tier real-world ladder - every task at that tier and below - where tier 1 is a localized single-file bug and tier 5 is a problem that took a human engineer hours. It solves some but not all tasks up at tier 5. Its capability score on real-world@v3 is 53% (grade D). Last benchmarked 2026-08-10.
Compare
- gpt-5.6-terra vs gemini-3.1-pro
- gpt-5.6-terra vs gemini-3.6-flash
- gpt-5.6-terra vs claude-fable-5
- gpt-5.6-terra vs claude-haiku-4-5
- gpt-5.6-terra vs claude-opus-5
- gpt-5.6-terra vs claude-sonnet-5
- gpt-5.6-terra vs gpt-5.6-luna
- gpt-5.6-terra vs gpt-5.6-sol
- gpt-5.6-terra vs composer-2.5
- gpt-5.6-terra vs grok-4.5
In the tools directory
- OpenAI Codex CLI - agentic-coding