unbiased ai

For agent builders

Agents do not pay per answer, they pay per everything-they-read. We published a real session ledger so this page can argue from receipts: 78x more tokens read than written. Here is what that means for your architecture and your bill.

Agentic workloads bill differently from chat: our measured Claude Code session read 144,929 context tokens to write 1,859, a 78x ratio, and cost $0.5887 for one bug fix. For agent builders the levers are context discipline, caching (2.6x measured), and per-request model selection, in that order.

The ledger, since nobody else publishes one

Five API calls: read the repo and run tests, read the buggy file, make two edits, re-run tests, summarize. Output tokens were 1.3% of the total moved; everything else was the agent reading. Caching turned what would have been a $1.54 session into $0.59. The full per-call table is in the Fable 5 breakdown, and it generalizes: whatever model your agent runs on, context handling dominates the bill.

The session ledger, interactive: flip caching off

#What the agent didContext inOutCost
1Read the repo, ran the tests27,261307$0.3561
2Read the buggy file28,202129$0.0455
3Fixed both bugs (two edits)28,8081,022$0.0869
4Re-ran the tests: all pass30,148157$0.0534
5Wrote the summary30,500244$0.0467
One bug fix, five calls144,9291,859$0.5887
$0.59 total. Call 1 pays the cache-write premium ($0.36) so calls 2–5 can reread 114k tokens at a tenth of the rate.

Metered on Claude Fable 5 rates: $10/M input, $50/M output, cache writes at 1.25× input, cache reads at 0.1×. The "without caching" column re-prices the identical token movement with every call re-sending full context at the input rate. Totals also include the session's 10 uncached fresh input tokens. Full narrative in the Fable 5 breakdown.

Three architecture consequences

Cache-stable prompts are a design requirement, not an optimization: system prompts and tool definitions that never churn keep reads at a tenth of the input rate. Context diets beat cheaper models: an agent that rereads half as much saves more than one that switches down a tier and starts failing. Per-step model selection matters most in agents: a session mixes trivial steps (run the tests) with hard ones (find the bug), and paying one flat frontier rate for both is exactly the waste a blended model eliminates per request.

Price your agent's loop

Context (reading)$0.4190
Output (writing)$0.1000
$0.5190 per run, $519 per 1,000 runs on Claude Fable 5. Context is 81% of this bill; the model choice is the second-order term.

Defaults are our measured session's averages (5 calls, ~29k context/call, 79% of context served from cache). Cache reads at 10% of input rate; the 1.25× cache-write premium is left out for simplicity (it was ~13% of our measured bill). Rates verified 2026-07-28.

Building on Pareto

Pareto 26.9’s published benchmark scores are on the model card. Measured task costs and a composite score have not been published for this release.

Context diet vs model downgrade: which actually saves more?

Today (Opus 5)$0.2595/run
A: context diet$0.1967/run
B: cheaper tier + retries$0.0908/run
At 15% failure, the downgrade to Claude Haiku 4.5 still wins ($0.0908 vs $0.1967 per run). Verify that failure rate on YOUR evals before trusting it.

Baseline: our measured session shape (5 steps, 29k context/step, 79% cached, 400 output/step) on Claude Opus 5 at $5/$25. Failed cheap runs are re-billed as a full baseline run on the frontier model, which is generous to Option B: review time is free here and is not in real life. Illustrative math with editable assumptions; the measured anchor is the 78x read ratio.

Prove it on your agent

Point a dev branch of your agent at Pareto, run your existing eval suite, and compare metered cost per completed task against your current setup. The $0.5887 session above is the Fable 5 bill; Pareto's rate card runs at a quarter of Fable's input rate and 15% of its output rate, so $100 in credits covers hundreds of sessions like it. If your agent framework needs config help, the FAQ covers the common setups.

Questions we get from this seat

Why is my agent so expensive to run?

Context reads. Our measured session moved 78x more tokens reading than writing; agents reread files, tool outputs, and history constantly. Cut rereads and cache stable context before touching the model choice.

Does prompt caching work mid-session for agents?

Yes, and it is where caching shines: session context accumulates and gets reread every call. Our measured session's cache reads billed at a tenth of input rate, saving 2.6x overall.

Can I pin specific agent steps to specific models?

With hand-rolled routing, yes, and it becomes a config file you maintain forever. A blended model does the same thing per request without the table; our benchmarks include the agentic suites so you can check the quality claim first.

Compare any two models

VS

List rates and dated competitor measurements: prices and measured bills. Pareto 26.9 measured task costs are not published. Verbosity: Edition 2.

Run your agent's evals against it. Session costs depend on your token use.
Get started with Unbiased