Codex
28 days of Tibo
Reviewed: · 25/28
Full reset ≠ banked reset · unknown ≠ no
America/Los_Angeles · · Original promise ↗ · Public record ↗
| Day (date) | Promise | Fact-checking | Passed? | Reset given? | Evidence and sources |
|---|---|---|---|---|---|
| Default speed ≈ +50% · Speed clarification · 30 → 50 tok/s | Timing definitions differ | Unresolved | No | › | |
| Free auto-review · API · meeting notes | Reset completion confirmed | Yes | Full reset | › | |
| Codex cloud · GPT-6 in Chat | Banked, not a completed reset | Unresolved | Banked reset | › | |
| Instant steering · Sol ultrafast | Improvement not verified | Unresolved | No | › | |
| Composer predictions · Mobile dot | Improvement not verified | Unresolved | No | › | |
| Full reset | Reset completion confirmed | Yes | Full reset | › | |
| Sol default context · 500k | Improvement not verified | Unresolved | No | › | |
| Full reset | Reset completion confirmed | Yes | Full reset | › | |
| App state on disk ≈ −60% | Improvement not verified | Unresolved | No | › | |
| Full reset | Reset completion confirmed | Yes | Full reset | › | |
| Fewer model-at-capacity errors | Improvement not verified | Unresolved | No | › | |
| More effective usage · context cache | Allowance effect unverified | Unresolved | No | › | |
| Full reset | Reset completion confirmed | Yes | Full reset | › | |
| Edit previous messages | Improvement not verified | Unresolved | No | › | |
| Cooked | Reset completion confirmed | Yes | Full reset | › | |
| API cache · 12h; Codex / Work · indefinite | Improvement not verified | Unresolved | No | › | |
| Full reset | Completion reported next day | Unresolved | Full reset | › | |
| Decisions API GA · performance updates | Completion reported next day | Unresolved | Full reset | › | |
| Faster Dot calls · reasoning cache fix · CLI update | Improvement not verified | Unresolved | No | › | |
| Full reset | Reset completion confirmed | Yes | Full reset | › | |
| Full reset | Reset completion confirmed by tracker | Yes | Full reset | › | |
| Terra is back · better writing and computer use · 2× usage | Improvement not verified | Unresolved | No | › | |
| Full reset | Reset completion confirmed | Yes | Full reset | › | |
| Text Dot directly · sharing, voice notes, Siri | Improvement not verified | Unresolved | No | › | |
| Usage limits doubled today · Pro 100/200/500, Business Premium | Improvement not verified | Unresolved | Unknown | › |
- Fulfilled
- 9
- Not fulfilled
- 0
- Partly fulfilled
- 0
- Unresolved
- 16
| Day (date) | Promise | Fact-checking | Passed? | Reset given? | My rating | Evidence and sources |
|---|---|---|---|---|---|---|
| Default speed ≈ +50% · Speed clarification · 30 → 50 tok/sMeasurements and chart ↗ | Timing definitions differ | Unresolved | No | Not rated | › | |
| Free auto-review · API · meeting notes | Reset completion confirmed | Yes | Full reset | Not rated | › | |
| Codex cloud · GPT-6 in Chat | Banked, not a completed reset | Unresolved | Banked reset | Not rated | › | |
| Instant steering · Sol ultrafast | Improvement not verified | Unresolved | No | Not rated | › | |
| Composer predictions · Mobile dot | Improvement not verified | Unresolved | No | Not rated | › | |
| Full reset | Reset completion confirmed | Yes | Full reset | Not rated | › | |
| Sol default context · 500k | Improvement not verified | Unresolved | No | Not rated | › | |
| Full reset | Reset completion confirmed | Yes | Full reset | Not rated | › | |
| App state on disk ≈ −60% | Improvement not verified | Unresolved | No | Not rated | › | |
| Full reset | Reset completion confirmed | Yes | Full reset | Not rated | › | |
| Fewer model-at-capacity errors | Improvement not verified | Unresolved | No | Not rated | › | |
| More effective usage · context cacheMeasurements and chart ↗ | Allowance effect unverified | Unresolved | No | Not rated | › | |
| Full reset | Reset completion confirmed | Yes | Full reset | Not rated | › | |
| Edit previous messages | Improvement not verified | Unresolved | No | Not rated | › | |
| Cooked | Reset completion confirmed | Yes | Full reset | Not rated | › | |
| API cache · 12h; Codex / Work · indefinite | Improvement not verified | Unresolved | No | Not rated | › | |
| Full reset | Completion reported next day | Unresolved | Full reset | Not rated | › | |
| Decisions API GA · performance updates | Completion reported next day | Unresolved | Full reset | Not rated | › | |
| Faster Dot calls · reasoning cache fix · CLI update | Improvement not verified | Unresolved | No | Not rated | › | |
| Full reset | Reset completion confirmed | Yes | Full reset | Not rated | › | |
| Full reset | Reset completion confirmed by tracker | Yes | Full reset | Not rated | › | |
| Terra is back · better writing and computer use · 2× usage | Improvement not verified | Unresolved | No | Not rated | › | |
| Full reset | Reset completion confirmed | Yes | Full reset | Not rated | › | |
| Text Dot directly · sharing, voice notes, Siri | Improvement not verified | Unresolved | No | Not rated | › | |
| Usage limits doubled today · Pro 100/200/500, Business Premium | Improvement not verified | Unresolved | Unknown | Not rated | › |
Day 1 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- No
Promise
- Default speed ≈ +50% ↗
Astra and Sol through Sign in with ChatGPT; default speed, no settings change.
- Speed clarification · 30 → 50 tok/s ↗
Criterion
Default Astra / Sol speed: approximately +50%, clarified as 30 → 50 TPS, using a comparable timing definition.
The 50 TPS declaration and our latency-inclusive response speed have no established common timing definition. Crossing the line alone proves neither outcome.
Sources
Day 2 · Evidence and sources
- Passed?
- Yes
- Reset given?
- Full reset
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The public record confirms completion of the full reset, satisfying the reset alternative.
Sources
Day 3 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- Banked reset
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
A banked reset is not a full reset that day. The announced improvement still needs verification against the daily criterion.
Sources
Day 4 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- No
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The announcement records what was promised. Evidence that it satisfies the daily improvement criterion has not been recorded.
Sources
Day 5 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- No
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The announcement records what was promised. Evidence that it satisfies the daily improvement criterion has not been recorded.
Sources
Day 6 · Evidence and sources
- Passed?
- Yes
- Reset given?
- Full reset
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The public record confirms completion of the full reset, satisfying the reset alternative.
Sources
Day 7 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- No
Promise
- Sol default context · 500k ↗
Default context: 272k → 500k; a manual override can still reach 1M.
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The announcement records what was promised. Evidence that it satisfies the daily improvement criterion has not been recorded.
Sources
Day 8 · Evidence and sources
- Passed?
- Yes
- Reset given?
- Full reset
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The public record confirms completion of the full reset, satisfying the reset alternative.
Sources
Day 9 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- No
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The announcement records what was promised. Evidence that it satisfies the daily improvement criterion has not been recorded.
Sources
Day 10 · Evidence and sources
- Passed?
- Yes
- Reset given?
- Full reset
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The public record confirms completion of the full reset, satisfying the reset alternative.
Sources
Day 11 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- No
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The announcement records what was promised. Evidence that it satisfies the daily improvement criterion has not been recorded.
Sources
Day 12 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- No
Promise
- More effective usage · context cache ↗
Context shared across subagents, longer cache retention and improved routing.
Criterion
More effective usage through context reuse and cache, measured on comparable work and settings.
A cache-read share does not establish increased effective allowance; a comparable allowance measurement is missing.
Sources
Day 13 · Evidence and sources
- Passed?
- Yes
- Reset given?
- Full reset
Promise
- Full reset ↗
Completion confirmed after the Day 13 reset announcement.
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The public record confirms completion of the full reset, satisfying the reset alternative.
Sources
Day 14 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- No
Promise
- Edit previous messages ↗
Edit earlier messages within a thread in the desktop app.
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The announcement records what was promised. Evidence that it satisfies the daily improvement criterion has not been recorded.
Sources
Day 15 · Evidence and sources
- Passed?
- Yes
- Reset given?
- Full reset
Promise
- Cooked ↗
The brief post does not establish a shipped improvement; a full reset was confirmed later that Pacific day.
- Full reset ↗
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The public record confirms completion of the full reset, satisfying the reset alternative.
Sources
Day 16 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- No
Promise
- API cache · 12h; Codex / Work · indefinite ↗
The announcement promises cache hits when signed-in users resume older conversations, without a return penalty.
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The announcement records what was promised. Evidence that it satisfies the daily improvement criterion has not been recorded.
Sources
Day 17 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- Full reset
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The public tracker records completion on the next Pacific day, without an X completion post. It does not establish completion within the promised day; the daily verdict remains unresolved.
Sources
Day 18 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- Full reset
Promise
- Decisions API GA · performance updates ↗
The post also mentions OpenResponses joining the Linux Foundation and Uber adopting Codex / Work; it announces a reset.
- Full reset announced ↗
- Reset completion reported next day ↗
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The public tracker records completion on the next Pacific day, without an X completion post. It does not establish completion within the promised day; the daily verdict remains unresolved.
Sources
Day 19 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- No
Promise
- Faster Dot calls · reasoning cache fix · CLI update ↗
The announcement says Dot calls connect faster and changing reasoning mid-conversation no longer invalidates cache in Codex / Work.
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The announcement records what was promised. Evidence that it satisfies the daily improvement criterion has not been recorded.
Sources
Day 20 · Evidence and sources
- Passed?
- Yes
- Reset given?
- Full reset
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The public record confirms completion of the full reset, satisfying the reset alternative.
Sources
Day 21 · Evidence and sources
- Passed?
- Yes
- Reset given?
- Full reset
Promise
- Full reset announced ↗
- Full reset · tracker confirmation ↗
The public tracker confirms completion within Day 21 in Los Angeles, without an X completion post.
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The public tracker records full-reset completion at 23:10 on Day 21 in Los Angeles, without an X completion post. This satisfies the same-day reset alternative.
Sources
Day 22 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- No
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The announcement records what was promised. Evidence that it satisfies the daily improvement criterion has not been recorded.
Sources
Day 23 · Evidence and sources
- Passed?
- Yes
- Reset given?
- Full reset
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The public record confirms completion of the full reset, satisfying the reset alternative.
Sources
Day 24 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- No
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The announcement records what was promised. Evidence that it satisfies the daily improvement criterion has not been recorded.
Sources
Day 25 · Evidence and sources
- Passed?
- Unresolved
- Reset given?
- Unknown
Promise
Criterion
One clear improvement relevant to most Codex / Work users, or a completed full reset that day.
The announcement records what was promised. Evidence that it satisfies the daily improvement criterion has not been recorded.
Sources
Day 26–28 · Awaiting review