Field notes ยท from three hands-on evals

Saving money with Fable 5

For agents Reading this with an AI agent? Point it at the markdown version instead: fable-5-news.pages.dev/savings.md

Fable 5 is the most capable Claude model - and, as a manager, still the priciest way to burn a Claude subscription. Three hands-on evals (a solo-vs-delegation orchestration test at n=2, a coding-subagent shootout, and a difficulty/domain follow-up) found exactly one place it wins outright - as a worker on open-ended, exploratory tasks - and confirm it loses everywhere else measured. Takeaways first, evidence below.

Updated Aug 5, 2026 for Claude Opus 5 - plus a correction.

Six Opus 5 cells were added across the orchestration and difficulty/domain evals. Headline: Opus 5 has the lowest per-token rates on this page but is the most volume-hungry model measured, so it costs more than Opus 4.8 in both roles where the two ran head to head. The newest model is not automatically the cheapest.

Separately, a bug in the S-Log3 objective checker has been fixed. It had produced a false "all models fail identically" result that this page reported as a model finding for several weeks. All four cells actually pass. Full correction in Evidence 3.

Same feature. Same workers. Different manager.

Claude spend to manage one real coding task through the same pool of cheap workers, across effort levels

Fable 5 manageshigh effort, avg of 2 $6.42
Opus 5 manageshigh effort, midpoint of 2 runs* $3.59–4.56
Opus 4.8 manageshigh effort, midpoint of 2 runs* $2.13–4.74
Sonnet 5 manages--effort high, single run $3.29
−26% to −67% Claude spend, with every quality check still passing in every cell. Swapping the manager off Fable saves money whichever Claude model you swap to - but the newest model isn't automatically the cheapest, and the spread within a single model is wide enough that you should not pick on a single run.

*Bars for the two Opus rows show the midpoint of their range, since a single-length bar can't draw a range; the labels give the actual span. Both Opus rows are two runs of the same configuration, not two settings. In one run of each pair the eval script passed --effort high explicitly; in the other it passed no --effort flag at all - and Anthropic's docs state that omitting the parameter is equivalent to high on both models. So these ranges measure run-to-run variance, not an effort discount: $2.13–4.74 for Opus 4.8 (a 2.2x spread) and $3.59–4.56 for Opus 5 (1.3x). Do not read the low end of either range as a setting you can choose. Opus 4.8's cheap run got there by burning 2.9x fewer total tokens (1.34M vs 3.88M) while its per-token rate was actually higher ($1.59/1M vs $1.22/1M) - it simply did less work that time. This is also the main reason every n=1 figure on this page is labelled as directional.

Sonnet 5 manages, high effort ($3.29, 1066s): same flat prompt, same external gpt/glm worker pool as every other bar on this chart - unlike the Sonnet-manager-with-Sonnet-workers confound check further down, this run isolates the manager-model swap alone. Fidelity held at 5 (median, 3 samples × 2 judges), same as every other cell.

All numbers from the three eval reports summarized below; small sample sizes throughout (n=1–3 per cell, except the orchestration eval's fidelity scores at 3 samples × 2 judges per cell).

Six ways to spend less

The evidence

"Solo," "flat," and "nested" are three ways of getting the same coding task done. The same real feature was built under each shape and measured on time, cost, and quality.

Opus 4.8 does all the work itself Solo

One strong Claude model works alone: reads the spec, writes the code, runs the tests, commits.

manager gpt gpt glm glm one hop: manager → workers Flat delegation

A Claude manager (Fable 5 or Opus) splits the task and hands pieces directly to cheaper workers on other subscriptions, then integrates and verifies.

Fable 5 Opus mgr Opus mgr glm glm gpt glm Nested delegation

Fable 5 delegates to Opus middle managers, who delegate to workers. Two hops between the spec and the code.

The workers throughout: GPT 5.5 (via codex, on a ChatGPT plan) and GLM 5.2 (via opencode, on a Z.AI plan) - so their token grind lands on those subscriptions, not the Claude plan.

How this was tested

One real feature (a preset pack/unpack command plus UI) was built under each orchestration shape, from frozen prompts in isolated git worktrees, with identical acceptance checks: the test suite, a typecheck, a byte-identical CLI round trip, and a UI render check. Every cell across ten runs passed all four. Fidelity was re-collected with a rubric-anchored method (3 samples × 2 judge models per diff, with two contested spec checks scored objectively by code inspection instead of judge opinion) after the original single-sample judging proved noisy. Solo (Opus 4.8) ran twice, solo Fable once, solo Opus 5 once, Fable-managed flat twice at high effort and once at low, Opus-4.8-managed flat twice at high effort (once with the flag explicit, once relying on the default), Opus-5-managed flat twice on the same basis, Sonnet-5-managed flat once at high effort with the same original gpt/glm worker pool, and nested twice. The three Opus 5 cells were added on Aug 5, 2026 and carry acceptance results but no rubric fidelity re-judge, so they are marked "not judged" below rather than assigned a score. A follow-up confound check reran the same flat task with the worker pool fixed to claude-sonnet-5 (via the Agent tool) under three different managers - Fable, Opus, and Sonnet 5 itself, all at high effort - isolating manager-model cost from cross-vendor delegation overhead.

Wall clock

Seconds to complete the feature, lower is better

Solo Opus 4.8, high effort (default), avg of 2614s
Solo Fable, high effort852s
Flat Opus 5 mgr, high effort (default, no flag)868s
Solo Opus 5, high effort (default)877s
Flat Opus 4.8 mgr, high effort (default, no flag)904s
Flat Fable manager, high effort, avg of 21044s
Flat Opus 4.8 mgr, high effort (explicit)1057s
Flat Sonnet 5 manager, high effort1066s
Nested Fable (--effort high) → Opus 4.8 x2 (no effort flag), avg of 21207s
Flat Fable manager, low effort1276s
Flat Opus 5 mgr, high effort (explicit)1347s

Every delegated shape ran workers concurrently and still lost to Opus 4.8 working alone - overhead beat parallelism at this task size. Solo alone doesn't guarantee a win, though: Fable-solo and Opus 5-solo were both slower than several delegated shapes. Opus 5 is the slowest manager measured at high effort (1347s) and among the faster ones with no effort flag (868s) - the same configuration on paper, since Opus 5 defaults to high effort, which puts a floor under how much of this chart is run-to-run noise.

Claude cost, management layer

Dollars of Claude usage; workers ride other plans, so this is not a total-system claim

Flat Opus 4.8 mgr, high effort (default, no flag)$2.13
Flat Sonnet 5 manager, high effort$3.29
Flat Opus 5 mgr, high effort (default, no flag)$3.59
Flat Opus 5 mgr, high effort (explicit)$4.56
Flat Opus 4.8 mgr, high effort (explicit)$4.74
Solo Opus 4.8, high effort (default)$4.96
Flat Fable manager, low effort$5.27
Flat Fable manager, high effort$6.42
Solo Opus 5, high effort (default)$6.73
Nested Fable --effort high → Opus 4.8 no effort flag, avg of 2$7.87
Solo Fable, high effort$12.77

Managing the worker pool still costs less than doing the work yourself at the same model's prices, for every Claude manager measured. Fable solo remains the single most expensive cell. The newest model does not automatically move you down this chart: Opus 5 solo ($6.73) is the second-priciest cell here and 36% above Opus 4.8 solo, while Opus 5 as manager ($3.59–4.56) lands between Sonnet 5 and Opus 4.8-high.

Effective rate

Dollars per million total tokens (incl. cache) - shows the tax is per-token, not per-task-volume

Flat Sonnet 5 mgr, high effort$0.61
Solo Opus 5, high effort (default)$0.78
Solo Opus 4.8, high effort (default), avg$0.80
Flat Opus 5 mgr, high effort (explicit)$0.86
Flat Opus 5 mgr, high effort (default, no flag)$1.05
Flat Opus 4.8 mgr, high effort (explicit)$1.22
Flat Opus 4.8 mgr, high effort (default, no flag)$1.59
Nested Fable --effort high → Opus no effort flag, avg$1.66
Solo Fable, high effort$1.93
Flat Fable mgr, low effort$2.17
Flat Fable mgr, high effort$2.47–2.71

Fable blends to $1.9–2.7/1M tokens regardless of effort; Opus blends to $0.78–1.6/1M, and Opus 5 sits at the bottom of that band in every role it ran. It's a rate tax, not a volume tax. But note what this chart can't tell you: Opus 5 has the best rate of any Opus cell here and still produced the second-priciest bill on the cost chart above. Rate only sets the price per token - the model decides how many to spend.

Confound check: manager model, worker pool held fixed

Same flat-delegation task, workers fixed to claude-sonnet-5 (via the Agent tool) so cross-vendor overhead can't explain the gap - total Claude-side spend, n=1 each

Sonnet 5 manager + Sonnet 5 workers, high effort$4.59
Opus manager + Sonnet 5 workers, high effort$5.37
Fable manager + Sonnet 5 workers, high effort$9.65

With external (GPT/GLM) workers, Fable-manager cost $6.42 because the worker grind rode other subscriptions. Swap those workers for Sonnet 5 and Fable-manager jumps to $9.65 - the workers' cost now lands on the same Claude plan as the manager. Sonnet-as-manager was also fastest (720s vs 855s Fable, 871s Opus) and cheapest per token ($0.64/1M vs Opus's $0.88/1M and Fable's $1.39/1M in this same-vendor stack). Fidelity held at 5/5 (median) for all three, rejudged the same rubric-anchored way as every other cell.

Show the full data table
CellWall clockClaude cost (mgmt)WorkersFidelity (rejudge median, 1–5)
Solo (Opus 4.8, high effort (default), avg of 2)614s$4.96none5
Solo (Fable, high effort, n=1)852s$12.77none5
Flat (Fable manager, avg of 2, high effort)1044s$6.422 gpt + 2 glm5
Flat (Fable manager, low effort, n=1)1276s$5.273 gpt + 3 glm5
Flat (Opus 4.8 manager, high effort - default, no flag, n=1)904s$2.132 gpt + 2 glm5
Flat (Opus 4.8 manager, high effort - explicit, n=1)1057s$4.742 gpt + 2 glm5
Solo (Opus 5, high effort - default, n=1)877s$6.73nonenot judged
Flat (Opus 5 manager, high effort - default, no flag, n=1)868s$3.595 gpt + 1 glmnot judged
Flat (Opus 5 manager, high effort - explicit, n=1)1347s$4.562 gpt + 4 glmnot judged
Flat (Sonnet 5 manager, high effort, n=1)1066s$3.291 gpt + 4 glm5
Nested (Fable --effort high → Opus x2, no effort flag, avg of 2)1207s$7.871–7 gpt/glm mix5
Flat (Fable manager, high effort, n=1) — Sonnet 5 workers855s$9.654 sonnet-55
Flat (Opus manager, high effort, n=1) — Sonnet 5 workers871s$5.374 sonnet-55
Flat (Sonnet 5 manager, high effort, n=1) — Sonnet 5 workers720s$4.594 sonnet-55

Plan burn per cell (all Claude layers): solo Opus ~80 turns / ~6.1M cache reads / ~39k output; solo Fable ~6.3M / ~43k; Fable-flat high ~44 turns / ~2.3M / ~30k; Fable-flat low ~2.25M / ~17.4k; Opus-flat, no effort flag, 31 turns / 1.2M / 22k; Opus-flat high ~3.63M / ~35.9k; nested avg ~26 turns / ~4.1M / ~72k; Fable/Opus/Sonnet-manager-with-Sonnet-workers cells ~6.1–7.1M combined tokens each, all high effort.

The three Sonnet-worker cells above are a separate confound check (see the "Confound check" chart), not directly comparable to the GPT/GLM-worker rows on Claude cost - their worker spend rides the same Claude plan as the manager, while GPT/GLM workers ride other subscriptions entirely.

The "solo is higher fidelity" gap did not survive a rubric-anchored re-judge.

The original blind judge (one sample per diff) scored solo 5/5 and every delegated run 4, but reading the diffs directly found the same two deviations (non-atomic unpack, no export-set selector) present in the solo runs too - just unflagged. A follow-up re-judge (3 samples × 2 judge models per diff, those two checks scored by code inspection instead of judge opinion) confirmed it: all ten cells land at median 5. Nothing was "lost in the delegation telephone game" except one verified nested-only defect - a worker with no project context hardcoded its sandbox path into a committed test, which the rubric's granularity doesn't register as a fidelity-score difference. Nested and both Opus-manager cells remain n=1–2, so treat magnitudes as provisional.

"Default effort" for Opus most likely means high effort, not a lower tier.

Cells labelled "default, no flag" never had --effort passed at all (solo Opus 4.8, one Opus 4.8-manager run, one Opus 5-manager run, solo Opus 5, and the nested run's Opus secondmates). This is not a lower effort tier. Anthropic's documentation states that omitting the effort parameter produces exactly the same behavior as setting it to high, on Opus 4.8 and Opus 5 alike - so each "default" run and its explicit --effort high twin were running the same nominal configuration.

That makes those two pairs the most useful noise measurement on this page, because the only thing separating each pair is the run itself. Opus 4.8: $2.13 vs $4.74, a 2.2x spread. Opus 5: $3.59 vs $4.56, 1.3x. The token accounting confirms the mechanism rather than an effort discount - Opus 4.8's $4.74 run had the lower per-token rate ($1.22/1M vs $1.59/1M) and cost more only because it burned 2.9x more total tokens (3.88M vs 1.34M, mostly cache reads from more turns). It simply did more work that time.

Two consequences. First, don't read the cheap end of either range as a setting you can reach for. Second, a 2.2x spread within one configuration is wider than most of the between-model gaps this page reports, so every n=1 comparison here should be read as directional only - which is why the effort-level conclusions rest on the per-token rate columns, where the pattern is stable, rather than on single-run dollar totals.

So where does Fable 5 fit?

Everywhere above, Fable loses on cost, speed, or both. Two things it's actually shown, with evidence:

Both points come from the same design task, so treat this as one coherent data point rather than two independent confirmations - but it now has an Opus 5 cell pointing the same way as the Opus 4.8 one, which is more than it had before. Everything else this page measured - management, solo work on a normal task, speed - Opus won outright; see the numbers above.

Untested: Fable 5 as a reviewer, not a doerno cell run for this

None of the three evals above tested this shape. It's a distinct role from everything measured so far: Opus (or the worker pool) still writes the diff; Fable 5 only reads spec + diff and judges it, then either approves or sends it back for changes. That's closer to the blind-judge role the eval's own fidelity scoring used (gpt-5.5 and GLM 5.2 as judges) than to Fable-as-manager or Fable-as-worker.

Opus / workers writes the diff spec + diff Fable 5 reviews reads only, doesn't write pass Ship fail → needs changes, revise
Why it might work
  • Review is a bounded-context task - reading a spec and a diff, not running tool loops or writing code. The eval's rate tax showed up on management and authoring volume; a review pass could burn far fewer tokens per call than either role tested.
  • The re-judge findings in Evidence 1 show single-sample, unanchored judging missed real defects (the two contested checks were absent everywhere, but only got flagged inconsistently) - a more capable reviewer model, used carefully with an anchored rubric, is exactly the kind of fix that closed that gap.
  • It reuses spend where Fable's extra capability is plausibly the bottleneck (catching subtle spec deviations) rather than where it's irrelevant (grinding through routine edits), which is the same logic behind "pay for capability where capability is scarce."
Why it might not
  • The rate tax measured in Evidence 1 is per-token, not volume-dependent - a small review call still runs at Fable's ~1.9-3.3x rate premium over Opus. Cheap in absolute dollars, but not rate-competitive with a cheaper reviewer (e.g. a smaller Claude model or one of the existing judge models).
  • A review-then-revise loop is another hop, and Evidence 1's nested-delegation cells found each hop adds wall clock and one class of real risk: a second-hop context loss (the worker that hardcoded its sandbox path). A review hop that sends work back for changes could reproduce that pattern rather than avoid it.
  • Better judgment doesn't fix a bad rubric. The eval's own single-sample judging noise came from unanchored scoring criteria, not judge model weakness (gpt-5.5 and GLM 5.2 agreed on every diff once rubric-anchored). A stronger reviewer model may not help if the failure mode is the checklist, not the checker.

Speculative This is reasoning from the existing evidence, not a new measurement - no cell in any of the three evals ran Fable 5 as a blind reviewer feeding back to Opus. It would need its own eval: same spec, Opus (or the worker pool) writes, Fable 5 reviews against a rubric, and the loop is timed and costed like every other cell on this page.