# Toast 1: what happens when you stop making the frontier model do the searching

> Satyajit Ghana — Head of Engineering @ Inkers Technology
> canonical: https://ai.thesatyajit.com/articles/toast-1
> date: 2026-08-14
> tags: retrieval, agents, rag, search, cost, explainer
[Toast 1](https://www.mixedbread.com/blog/toast-1) is Mixedbread's first specialised search agent, released 2026-08-13. The pitch is division of labour: instead of a frontier model burning its context window navigating a corpus, Toast 1 takes the whole search loop — decomposes the query into subqueries, gathers evidence, inspects sources, curates what matters — and hands back a package. The frontier model spends its tokens reasoning instead.

It runs standalone or as a subagent, and it is backend-agnostic: co-designed with Mixedbread Search but able to run over an existing index.

## The result worth reading twice

<TokenLadder />

Harvey's LAB firm-knowledge benchmark, on a 33-task subset, with **one model and one evaluation** and only the retrieval stack changing between runs:

| configuration | tokens | turns/task | score |
|---|---|---|---|
| vanilla agent | 80.6M | 21.7 | 55 |
| + Mixedbread Search | 47.0M | 14.6 | 55 |
| + Toast 1 subagent | 23.0M | 11.2 | 55 |

The score is the finding. It does not move. If quality had gone up you would be looking at a better agent; because it is identical across all three rows, what the experiment demonstrates is that **57.6M of the vanilla agent's 80.6M tokens were not contributing to the answer**. They were the cost of looking.

I recomputed the deltas and they reproduce: 47.0/80.6 is −41.7% against a stated −42%, 23.0/47.0 is −51.1% against −51%, and 80.6/23.0 is 3.50× against a stated 3.5×.

Two caveats belong right next to that, and Mixedbread states the first itself in a footnote: this is a **randomly selected 33-task subset**, chosen "to make repeated comparative runs tractable." And a score that lands on exactly 55 three times is coarse enough that a small quality change would not necessarily show up in it. The token reduction is a much more precisely measured quantity than the quality preservation it is paired with.

## The headline benchmark, and whose numbers they are

<Figure
  src="/articles/toast-1/fig1.png"
  alt="Scatter plot of answer correctness against cost per rollout on OfficeQA Pro V2, with cost on a log scale. Points for GPT-5.6 Luna, Terra and Sol, Claude Fable 5, Kimi K3, GLM 5.2 and Sonnet 5 appear both bare and inside Codex or Claude Code harnesses, along with Databricks Genie. Two points labelled Codex plus Toast 1 sit above and to the left of the previous Pareto frontier."
  caption="Answer correctness against cost per rollout on OfficeQA Pro V2. Genie and harness numbers are as reported by Databricks; the Codex + Toast 1 runs are Mixedbread's own. (mixedbread.com, Toast 1 launch post.)"
/>

On [OfficeQA Pro V2](https://www.mixedbread.com/blog/toast-1) — 90 questions on enterprise financial situations, released by Databricks — GPT-5.6 Sol running in Codex with Toast 1 as a subagent reaches **70% correctness at about $1.15 per task**. The previous best in Databricks' evaluation, Claude Fable 5 on Databricks Genie, was 60% at roughly $4.

The comparison that carries the most information is the one against itself: **GPT-5.6 Sol in Codex without Toast 1 reaches 33%**. Same model, same harness, and correctness doubles when the search loop is delegated.

The chart's own footnote is the thing to hold onto: "Genie and harness numbers as reported by Databricks; Codex + Toast 1 runs are ours." Half the points come from the benchmark's authors and half from the vendor being evaluated. That is a normal and disclosed arrangement, and it is still a different evidential status than a single evaluator running everything.

## As a standalone retriever

<Figure
  src="/articles/toast-1/fig2.png"
  alt="Scatter plot of NDCG at 10 against cost per query on BrowseComp Plus, log-scale cost. Each model shows a short line for its reasoning sweep. Toast 1 with RRF times 3 sits at about 0.86 NDCG for roughly $0.05 to $0.09 per query, higher than GPT-5.6 Terra, Opus 5, Kimi K3, Qwen, GLM, Sonnet 5, DeepSeek and Haiku 4.5, and level with GPT-5.6 Sol which costs several times more."
  caption="BrowseComp Plus: retrieval quality against cost per query, both axes measured. Toast 1's fusion configuration sits level with GPT-5.6 Sol at a fraction of the cost. (mixedbread.com, Toast 1 launch post.)"
/>

Evaluated as a retriever rather than a subagent — BrowseComp Plus, OfficeQA Pro and LongSeal, scored by NDCG@10 — Toast 1's fusion configuration lands in the same band as GPT-5.6 Sol and above Kimi K3, GLM, Opus 5 and Sonnet 5, while sitting an order of magnitude to the left on cost.

The chart shows something the prose does not dwell on: every other system is drawn as a *sweep*, a short line tracing what more reasoning effort buys. Toast 1's line is short and nearly flat. Whatever it is doing, spending more on it does not move quality much — which is the expected shape for a specialised model that is already doing the one thing it was trained for.

## The economics

<QueryEconomics />

A standard run is **$0.016–$0.023 per query at an eight-second median**; the fusion configuration is **$0.05–$0.07 at eleven seconds**. Token pricing is $0.30/M input, $0.04/M cached input with free cache writes, and $0.80/M output.

Against the frontier retrieval agents in the same evaluation, Mixedbread claims 7–11× cheaper. That multiple is the only cost figure given for the comparison group, so the band in the diagram above is their claim inverted rather than a published measurement — worth flagging, because the latency comparison beside it needs no such inference: **20 seconds to four minutes**, quoted directly.

An eight-second search subagent and a four-minute one are different products before price enters the discussion. If a frontier model is going to call search several times per task, the difference compounds into whether the task is interactive at all.

## What is not disclosed

No architecture. No parameter count. No training details, data, or method. No information about what Toast 1 is beyond what it does and what it costs. This is a product launch, not a model release, and every number in it is a system-level measurement.

That matters for one specific reason: the headline results are all **system** results. "GPT-5.6 Sol + Toast 1 in Codex reaches 70%" is a claim about a pipeline with at least three moving parts, and the contribution of each is not separable from the published data. The Harvey ladder is the closest thing to a controlled experiment on offer, and it is the one I would weight most — one model, one task set, one evaluator, one variable.

Mixedbread's own footnote places this alongside SID-1 and Chroma's Context-1 as a growing category of specialised search agents. That framing is right, and it is the more interesting story than any single benchmark: the bet is that retrieval is a distinct enough skill to be worth a dedicated model, and that the frontier model's context window is too expensive to spend on navigation.

The Harvey numbers are the strongest evidence for that bet I have seen stated plainly. Three quarters of a vanilla agent's tokens went to finding things, and removing that cost changed nothing about the answers.