CellCog’s Agent Max and Creative Max tiers now run on Claude Fable 5.2. The change went live at 01:29 UTC on November 11, 2026, two days after Anthropic released the model. Every Flash and Core tier, and Team Max, stays on Claude Opus 5.5. There is nothing to do on your side: a chat or an AI employee on Agent Max or Creative Max is already on the new model.
This was not a day-one move. When Fable 5.2 came out on November 9, no CellCog tier moved, and our Fable 5.2 release record said so. We ran it on three of our own AI employees on November 9 and 10. That was ordinary work and not a benchmark, so we have no numbers from it to publish. The two tiers moved on November 11.
On this page · 6 sectionsOpen
Since November 11, 2026, CellCog’s Agent Max and Creative Max tiers run on Claude Fable 5.2. Every Flash and Core tier, and Team Max, stays on Claude Opus 5.5.
This was not a day-one move. Anthropic released Fable 5.2 on November 9, 2026. We ran it on three of our own AI employees first, and the two tiers moved two days after the release.
Core and Max now run different models on Agent and on Creative. Until this change, every tier ran on Opus 5.5.
A switch between Core and Max on Agent or Creative now changes the model. The new model reads the conversation again from the start, so the first turn after the switch uses more credits than a normal turn.
Max work on Agent and Creative uses credits faster than before. Fable 5.2’s list price is twice that of Opus 5.5. We have not measured the real difference on our own work yet.
Nothing to do on your side. A chat or an AI employee on Agent Max or Creative Max is already on Fable 5.2, and its memory, role and task board carried over.
§ 01What changed
| Tier | Agent | Creative | Team |
|---|---|---|---|
| Flash | Opus 5.5 | Opus 5.5 | Opus 5.5 |
| Core | Opus 5.5 | Opus 5.5 | Opus 5.5 |
| Max | Fable 5.2 (was Opus 5.5) | Fable 5.2 (was Opus 5.5) | Opus 5.5 |
Two cells changed. Before this, every tier in the table ran on Opus 5.5. Core and Max moved to it on the day it was released, and Flash followed later that month.
AI employees follow their tier. An AI employee runs in the mode and tier it is set to. One set to Agent Max or Creative Max now runs on Fable 5.2, and every other AI employee runs on Opus 5.5.
You still do not pick a model. You pick a mode and a tier. Flash, Core, Max explains what each one is for.
Team Max did not move. Team is several agents that check each other’s findings, and it runs on Opus 5.5 at every tier.
§ 02What Fable 5.2 is
Anthropic released Claude Fable 5.2 on November 9, 2026 (launch page, release notes). Its guidance is to start on Opus 5.5 and move up to Fable for the hardest work. CellCog’s tiers now have that shape on Agent and Creative: Opus 5.5 on Flash and Core, Fable 5.2 on Max.
| Measure | Fable 5.2 | Opus 5.5 |
|---|---|---|
| FrontierCode 2.0 | 49.1% | 45.5% |
| Remote Labor Index | 41.3% | 21.3% |
| Input price | $8 | $4 |
| Output price | $40 | $20 |
These are Anthropic’s numbers. The Remote Labor Index row has a second source: Scale Labs listed the two models at 41.30 and 21.25 on its own leaderboard when we read it on November 10. We have not run CellCog on that index, and our record of it says what it does and does not measure. The table also shows the cost: Fable 5.2’s list price is twice that of Opus 5.5.
§ 03What changes for you
A switch between Core and Max now changes the model. On Agent and on Creative, Core runs on Opus 5.5 and Max runs on Fable 5.2. When you switch a running chat between them, the new model reads the conversation again from the start. The first turn after the switch therefore uses more credits than a normal turn. A switch between Agent Max and Creative Max keeps the same model, and so does a switch between Flash and Core.
Max work on Agent and Creative uses credits faster than before. CellCog bills by usage, and Fable 5.2 costs twice as much per token as Opus 5.5 at list price. We have not measured the real difference on our own work yet, so there is no figure here. When we have measured it, this page will say what we found. You pay for the work, not the hire, and the cost depends purely on how much work you assign. The plans are on pricing.
Nothing to set up. There is no model picker and no migration step. An AI employee’s memory, role and task board carried over, because none of that is stored in the model.
§ 04What we do not know yet
- How much faster Max uses credits. Not measured yet.
- How CellCog scores on Fable 5.2. Our October 2026 DeepResearch Bench run used Creative Max on Opus 5.5 (benchmarks). We have not run it again on Fable 5.2, so that result says nothing about the new model.
- How it holds up over many working days. The two tiers have been on it for hours. Three of our own AI employees have been on it for one to two days.
§ 05What we got wrong on the day
After the change, pages on this blog still said that every CellCog tier ran on Opus 5.5, or that no tier ran on Fable 5.2. We have found and fixed nine so far, the last of them 1 hour 55 minutes after the change. Three of the nine said it in a correction dated the same day, and those three carry a second correction line. We are still checking the rest of the blog. News records keep the line they had on their date.
§ 06Sources
- Anthropic, Claude Fable 5.2 and Mythos 5.2, November 9, 2026
- Anthropic, release notes and what is new in Fable 5.2
- Scale Labs, Remote Labor Index leaderboard, read November 10, 2026
- CellCog’s production configuration and worker log, read at 02:51 UTC on November 11, 2026
Q1Which CellCog tiers run on Claude Fable 5.2?
Two: Agent Max and Creative Max, since November 11, 2026. Agent Flash, Agent Core, Creative Flash, Creative Core and all three Team tiers run on Claude Opus 5.5.
Q2Does my AI employee run on Fable 5.2?
It does if it is set to Agent Max or Creative Max. An AI employee follows the mode and tier it is set to, so every other AI employee runs on Opus 5.5.
Q3Does this change what I pay?
Max work on Agent and Creative uses credits faster than before, because CellCog bills by usage and Fable 5.2’s list price is twice that of Opus 5.5. We have not measured the real difference on our own work, so we publish no figure yet. Flash, Core and Team work is unaffected. You pay for the work, not the hire, and the cost depends purely on how much work you assign.
Q4What happens when I switch a chat between Core and Max?
On Agent and on Creative the model now changes with the tier. The new model reads the conversation again from the start, so the first turn after the switch uses more credits than a normal turn. A switch between Agent Max and Creative Max keeps the same model. On Team, Core and Max run the same model.
Q5Did CellCog move to Fable 5.2 on release day?
No. On November 9, 2026 no tier moved, and our release record said so. We ran Fable 5.2 on three of our own AI employees on November 9 and 10, and the two tiers moved on November 11.