---
title: "Introducing Toast 1"
description: "Meet Toast 1, Mixedbread's search agent for knowledge-intensive tasks, matching or outperforming Claude Opus 5 and GPT-5.6 Sol while being up to 10× cheaper and 12× faster."
date: 2026-08-13
authors:
  - Mixedbread Team
tags:
  - Research
  - Product
tag: blog
url: /blog/toast-1
featured: true
---

**Toast 1**, our first specialised search agent, is available today. It provides frontier search quality, matching or outperforming Claude Opus 5 and GPT-5.6 Sol while being up to 10× cheaper and 12× faster. It performs best with Mixedbread Search, but it can work with any search backend.

Today, frontier models are now able to perform real knowledge work. They can reason, analyse, and find information in complex document collections. But they are also the most expensive models in the stack. As intelligence is increasingly metered, the need for specialised agents able to match their capabilities at a fraction of the cost is greater than ever.

Toast 1 can run as a standalone specialized retrieval agent, or as one of many subagents your frontier model already knows how to rely on. It fully takes over the search loop: given an initial query, it decomposes it into subqueries, gathers evidence, inspects sources, and curates the relevant context before returning it. This lets your agent spend its context and compute on the task that requires a generalist, frontier-level model: reasoning, acting, and producing the final answers.

> **Figure**: Waterfall trace of a Toast 1 agentic search: 16 tool calls across 3 rounds answering "How did the employment rate change in retail compares to the healthcare sector?" in 5.33 seconds (102,536 prompt tokens, 1,248 completion tokens).

| Start | Duration | Tool | Call | Result |
| --- | --- | --- | --- | --- |
| 0.00s | 159ms | inspect_metadata | Samples the store's metadata fields to learn which attributes can be filtered or ranked on. This store has none, so the agent relies on content search alone. | 0 metadata fields found |
| 0.00s | 159ms | search_corpus | How did the employment rate change in retail compares to the healthcare sector? | bootstrap search with the raw query, 5 chunks added to context |
| 0.16s | 992ms | thinking | The initial results discuss skill shifts but do not establish the requested employment-rate comparison. I will retrieve evidence on both sectors, plus exact phrasing and any chart or table that directly reports the rates. | round 1 → queued 5 parallel tool calls |
| 1.16s | 444ms | search_corpus | How did employment rates change in retail compared with healthcare? | 5 new chunks, 14 duplicates skipped |
| 1.16s | 445ms | search_corpus | What were the employment growth rates for retail and healthcare? | 5 new chunks, 7 duplicates skipped |
| 1.16s | 445ms | search_corpus | Which sectors saw rising or falling employment, including retail and healthcare? | 5 new chunks, 2 duplicates skipped |
| 1.16s | 445ms | grep | employment rate\|employment.*retail\|retail.*employment\|healthcare.*employment\|health care.*employment | lexical match over text and OCR, 16 matching chunks |
| 1.16s | 445ms | overview_search | How did retail and healthcare employment rates change relative to each other? | document-level overview pass, 25 chunks surfaced |
| 1.62s | 1502ms | thinking | The retrieved chunks now establish the key directions: healthcare employment rose, while retail employment was roughly flat to slightly declining, but the exact chart points still need clearer verification and the comparison should be quantified for both the United States and Western Europe. I will retrieve the full retail and healthcare pages and search for any explicit rate or trend statement. | round 2 → queued 5 parallel tool calls |
| 3.12s | 181ms | get_chunks | Pulls five previously seen chunks back into context for a closer read of the retail and healthcare chart pages. | 5 chunks restored |
| 3.12s | 181ms | search_corpus | How did U.S. retail employment change from 2000 to 2016, and how did healthcare employment change over the same period? | 5 new chunks, 10 duplicates skipped |
| 3.13s | 181ms | search_corpus | What were the employment trends for retail and healthcare in the United States and Western Europe? | 5 new chunks, 20 duplicates skipped |
| 3.13s | 180ms | search_corpus | Did retail employment fall while healthcare employment rose between 2000 and 2016? | 5 new chunks, 19 duplicates skipped |
| 3.13s | 179ms | grep | 16\.6\|15\.7\|20\.1\|19\.1\|retail.*employment\|employment.*retail\|healthcare.*employment\|employment.*healthcare | verifying exact chart figures, 16 matching chunks |
| 3.32s | 2010ms | thinking | The evidence now fully supports the comparison: healthcare employment rose substantially, while retail employment was essentially flat in the United States and fell in Western Europe. I will rank the direct chart chunks first, then the corroborating trend and occupational-shift chunks, and I will state the percentage changes computed from the chart endpoints. | round 3 → submits the final ranking |
| 5.33s | 0ms | submit_ranking | I prioritized chunks containing the 2000 and 2016 employment figures for retail and healthcare, then chunks explicitly stating the sectors' employment directions, followed by occupational and skill-shift support. I computed approximate percentage changes using the chart endpoints: U.S. retail rose from 15.6 to 16.6 million FTEs (+6.4%), Western European retail fell from 17.0 to 15.7 million (−7.6%); U.S. healthcare rose from 16.4 to 20.1 million (+22.6%), and Western European healthcare rose from 15.1 to 19.1 million (+26.5%). | 7 chunks returned, ranked by relevance |

## Pareto Optimal Search

This specialisation of agentic labor results in considerably cheaper search, but also in better end-to-end results on many realistic tasks. We found that Toast 1 establishes a new Pareto frontier across agentic workloads across cost per task and speed per task.

### Financial Analysis: OfficeQA Pro V2

OfficeQA Pro V2, [released by Databricks](https://www.databricks.com/blog/introducing-officeqa-pro-v2-new-benchmark-enterprise-grounded-reasoning), evaluates answer correctness across 90 questions in realistic, complex enterprise financial situations.

GPT‑5.6 Sol with Toast 1 made available as a sub-agent within Codex reaches 70% answer correctness at approximately \$1.15 per task: that is the highest score among the systems evaluated by Databricks in the OfficeQA v2 release, establishing new state-of-the-art performance in both quality and efficiency.

> **Figure**: Cost–quality Pareto on OfficeQA Pro V2: answer correctness versus cost per rollout (USD, log scale). Databricks Genie and model-provider harness numbers as reported by Databricks; Codex + Toast 1 runs are Mixedbread's.

| Model | Harness | Answer correctness | Cost / rollout | On Databricks' Pareto frontier |
| --- | --- | --- | --- | --- |
| GPT-5.6 Sol High | Codex + Toast 1 | 70% | $1.18 | no |
| GPT-5.6 Sol Low | Codex + Toast 1 | 61% | $0.75 | no |
| Claude Fable 5 | Databricks Genie | 60% | $4.26 | yes |
| GPT-5.6 Sol | Databricks Genie | 58.6% | $8.17 | no |
| GPT-5.6 Terra | Databricks Genie | 53.4% | $1.67 | yes |
| Claude Fable 5 | Model provider harness | 45% | $38.00 | no |
| Kimi K3 | Databricks Genie | 40% | $2.35 | no |
| GLM 5.2 | Databricks Genie | 40% | $3.13 | no |
| Sonnet 5 | Databricks Genie | 40% | $5.51 | no |
| GPT-5.6 Luna | Databricks Genie | 38.8% | $0.60 | yes |
| GPT-5.6 Sol | Model provider harness | 33.3% | $4.68 | no |
| GPT-5.6 Terra | Model provider harness | 21.2% | $1.26 | no |
| Sonnet 5 | Model provider harness | 15.7% | $5.00 | no |
| GPT-5.6 Luna | Model provider harness | 15% | $0.12 | yes |

By comparison, the previous best performer, Claude Fable 5 on Databricks Genie, reaches 60% correctness at approximately \$4 per task, while GPT-5.6 Sol within Codex without Toast 1 only reaches 33% correctness.

This improvement stems from reformulating the economics of evidence gathering. Toast 1's specialization allows it to produce high-quality, token-efficient evidence packages, leaving ample resources for the reasoning process to reach the final answer.

### Legal Agentic Benchmark - Firm Knowledge

[Harvey LAB's Law Firm Knowledge benchmark](https://www.harvey.ai/blog/legal-agent-bench-law-firm-knowledge) seeks to evaluate how well an agent can search and use institutional legal knowledge at large, realistic scales.

Legal work, by nature, is context-heavy. You cannot outargue someone with access to better, more relevant precedents and details. But it is also noisy: many situations are similar but vary by simple details, making it tricky to collect high quality evidence packages without numerous false positives.

On a randomly selected subset of 33 tasks,[^1] we found that GPT-5.6 Sol's answer quality remained constant across search methods.

> **Figure**: Total tokens on the Harvey LAB Law Firm Knowledge benchmark (33-task subset, GPT-5.6 Sol). All configurations reach the identical task score of 55; only the retrieval stack changes.

| Configuration | Tokens (benchmark total) | Turns / task | Task score |
| --- | --- | --- | --- |
| Vanilla agent | 80.6M (100% of vanilla) | 21.7 | 55 |
| + Mixedbread Search | 47.0M (58% of vanilla) | 14.6 | 55 |
| + Toast 1 subagent | 23.0M (29% of vanilla) | 11.2 | 55 |

Overall: 3.5× fewer tokens than the vanilla agent at the same score. Each step down is a reduction over the previous configuration: + Mixedbread Search uses 42% fewer tokens than Vanilla agent; + Toast 1 subagent uses 51% fewer tokens than + Mixedbread Search.

[^1]: We evaluated a randomly selected subset of 33 tasks to make repeated comparative runs tractable. Every configuration used the same tasks and evaluation setup; only the retrieval stack changed.

However, increasing search quality drastically increased token efficiency: replacing the vanilla agent's filesystem search with Mixedbread Search cut token usage from 80.6M to 47M at an identical task score. Subsequently adding Toast 1 as its dedicated search subagent reduced it further to 23M, and allowed it to finish in half the turns required by vanilla agent.

The introduction of a Mixedbread Search-powered Toast 1 preserved answer quality, while consuming 3.5× fewer tokens, leading to a cost reduction of over 60%. Toast 1 frees up the context window of frontier models to let them spend their tokens on reaching the right answer.

## Demo: Dig Deep Into Dwarkesh's Podcast

Benchmarks and numbers can only tell one part of the story. To truly understand how Toast 1 works, there is no better way than watching it search in action. At Mixedbread, we really enjoy [Dwarkesh's podcast](https://www.dwarkesh.com/), and thought being able to search deep into its transcripts would be fun.

You can try it yourself [here](https://dwarkesh-search-demo.vercel.app).

## Frontier Class Retrieval

Although it is a capable subagent for complex tasks, Toast 1 is also a capable standalone model, trained specifically for deep search. It represents the next step of our co-design approach behind our embedding models and Silo: the model, agent harness, and retrieval primitives are designed to work together.[^2]

> **Figure**: Retrieval quality versus cost and latency per query on BrowseComp Plus, OfficeQA Pro and LongSeal. Cost per query at list prices with prompt caching; latency is p50 per query. Rows without a cost had negative cost artifacts (cached tokens exceeding input) and are reported by latency only.

**BrowseComp Plus — retrieval quality vs. cost and latency per query**

| Model | Reasoning effort | NDCG@10 | Cost / query | p50 latency |
| --- | --- | --- | --- | --- |
| Toast 1 | default | 0.8046 | $0.0442 | 8s |
| Toast 1 | RRF ×3 | 0.8574 | $0.0991 | 10s |
| GPT-5.6 Sol | none | 0.7748 | $0.3599 | 66.6s |
| GPT-5.6 Sol | low | 0.8117 | $0.3657 | 82.9s |
| GPT-5.6 Sol | medium | 0.8365 | $0.3460 | 110.9s |
| GPT-5.6 Sol | high | 0.8632 | $0.4342 | 158.6s |
| GPT-5.6 Terra | none | 0.6883 | — | 55.1s |
| GPT-5.6 Terra | low | 0.7816 | $0.0491 | 86.6s |
| GPT-5.6 Terra | medium | 0.7577 | $0.1205 | 95.7s |
| GPT-5.6 Terra | high | 0.7820 | $0.1367 | 119.7s |
| GPT-5.6 Luna | none | 0.6388 | $0.0170 | 22.4s |
| GPT-5.6 Luna | low | 0.6455 | $0.0133 | 22s |
| GPT-5.6 Luna | medium | 0.7148 | $0.0161 | 27.5s |
| GPT-5.6 Luna | high | 0.7344 | $0.0168 | 41.2s |
| Opus 5 | off | 0.7330 | $0.5218 | 30.2s |
| Opus 5 | low | 0.6560 | $0.4615 | 23.3s |
| Opus 5 | medium | 0.7430 | $0.4239 | 28.1s |
| Opus 5 | high | 0.7890 | $0.5736 | 39s |
| Sonnet 5 | off | 0.5730 | $0.2676 | 31.4s |
| Sonnet 5 | low | 0.5080 | $0.2217 | 29.9s |
| Sonnet 5 | medium | 0.5730 | $0.2730 | 38.9s |
| Sonnet 5 | high | 0.6200 | $0.3475 | 51.6s |
| Haiku 4.5 | off | 0.3070 | $0.0961 | 24.3s |
| Haiku 4.5 | low | 0.3590 | $0.1308 | 46.4s |
| Haiku 4.5 | medium | 0.3960 | $0.1468 | 52.3s |
| Haiku 4.5 | high | 0.3780 | $0.1682 | 57.5s |
| GLM | high | 0.6480 | $0.1593 | 49.8s |
| GLM | xhigh | 0.6590 | $0.1882 | 111s |
| Qwen | low | 0.5840 | $0.1864 | 63.1s |
| Qwen | high | 0.6700 | $0.2413 | 108.4s |
| Qwen | xhigh | 0.6560 | $0.2506 | 112.4s |
| Kimi K3 | low | 0.5760 | $0.1690 | 39.5s |
| Kimi K3 | high | 0.6870 | $0.3125 | 84.5s |
| Kimi K3 | max | 0.7230 | $0.4069 | 149.6s |
| DeepSeek | off | 0.4740 | $0.0167 | 24.4s |
| DeepSeek | high | 0.6520 | $0.0226 | 73.6s |
| DeepSeek | max | 0.6520 | $0.0230 | 97.7s |

**OfficeQA Pro — retrieval quality vs. cost and latency per query**

| Model | Reasoning effort | NDCG@10 | Cost / query | p50 latency |
| --- | --- | --- | --- | --- |
| Toast 1 | default | 0.4905 | $0.0354 | 8s |
| Toast 1 | RRF ×3 | 0.5310 | $0.0794 | 10s |
| GPT-5.6 Sol | none | 0.5519 | $0.3375 | 70.7s |
| GPT-5.6 Sol | low | 0.5468 | $0.4002 | 64.9s |
| GPT-5.6 Sol | medium | 0.5640 | $0.3879 | 86.1s |
| GPT-5.6 Sol | high | 0.5731 | $0.4223 | 120.1s |
| GPT-5.6 Terra | none | 0.4600 | $0.0514 | 61.9s |
| GPT-5.6 Terra | low | 0.5225 | $0.1251 | 76.1s |
| GPT-5.6 Terra | medium | 0.5411 | $0.1351 | 96.9s |
| GPT-5.6 Terra | high | 0.5407 | $0.1530 | 114.7s |
| GPT-5.6 Luna | none | 0.4457 | $0.0095 | 75.8s |
| GPT-5.6 Luna | low | 0.4772 | $0.0142 | 54.8s |
| GPT-5.6 Luna | medium | 0.4694 | $0.0179 | 75.1s |
| GPT-5.6 Luna | high | 0.4984 | $0.0188 | 143.6s |
| Opus 5 | off | 0.4720 | $0.6501 | 25.2s |
| Opus 5 | low | 0.4620 | $0.4325 | 19s |
| Opus 5 | medium | 0.4790 | $0.5803 | 30.4s |
| Opus 5 | high | 0.4920 | $0.5345 | 42.3s |
| Sonnet 5 | off | 0.3930 | $0.2492 | 26.5s |
| Sonnet 5 | low | 0.3740 | $0.1568 | 14.5s |
| Sonnet 5 | medium | 0.4090 | $0.2577 | 33.3s |
| Sonnet 5 | high | 0.4240 | $0.3283 | 52.5s |
| Haiku 4.5 | off | 0.3260 | $0.0941 | 20.6s |
| Haiku 4.5 | low | 0.3630 | $0.1319 | 40.3s |
| Haiku 4.5 | medium | 0.3350 | $0.1129 | 50.9s |
| Haiku 4.5 | high | 0.3440 | $0.1514 | 59.6s |
| GLM | high | 0.4600 | $0.1567 | 76.7s |
| GLM | xhigh | 0.4120 | $0.2181 | 231.7s |
| Qwen | low | 0.3880 | $0.2478 | 85.2s |
| Qwen | high | 0.4160 | $0.2462 | 145.4s |
| Qwen | xhigh | 0.4340 | $0.2687 | 147.7s |
| Kimi K3 | low | 0.4020 | $0.2101 | 37.5s |
| Kimi K3 | high | 0.4590 | $0.3033 | 75.6s |
| Kimi K3 | max | 0.4730 | $0.3534 | 157.1s |
| DeepSeek | off | 0.4240 | $0.0186 | 28.8s |
| DeepSeek | high | 0.4240 | $0.0242 | 82.6s |
| DeepSeek | max | 0.4530 | $0.0256 | 81.7s |

**LongSeal — retrieval quality vs. cost and latency per query**

| Model | Reasoning effort | NDCG@10 | Cost / query | p50 latency |
| --- | --- | --- | --- | --- |
| Toast 1 | default | 0.6820 | $0.0237 | 6.6s |
| Toast 1 | RRF ×3 | 0.7080 | $0.0712 | 7.4s |
| GPT-5.6 Sol | none | 0.5490 | $0.0928 | 6.5s |
| GPT-5.6 Sol | low | 0.6170 | $0.1157 | 10.7s |
| GPT-5.6 Sol | medium | 0.5930 | $0.1401 | 11.4s |
| GPT-5.6 Sol | high | 0.6010 | $0.1553 | 12.3s |
| GPT-5.6 Terra | none | 0.4840 | $0.0270 | 5.6s |
| GPT-5.6 Terra | low | 0.4960 | $0.0504 | 8.2s |
| GPT-5.6 Terra | medium | 0.4860 | $0.0488 | 8.4s |
| GPT-5.6 Terra | high | 0.4670 | $0.0521 | 8.6s |
| GPT-5.6 Luna | none | 0.5150 | $0.0033 | 5.6s |
| GPT-5.6 Luna | low | 0.5350 | $0.0051 | 8.7s |
| GPT-5.6 Luna | medium | 0.5300 | $0.0061 | 9.5s |
| GPT-5.6 Luna | high | 0.5140 | $0.0073 | 11.4s |
| Opus 5 | off | 0.7770 | $0.4797 | 22.7s |
| Opus 5 | low | 0.7360 | $0.2809 | 13.9s |
| Opus 5 | medium | 0.7460 | $0.4093 | 19.1s |
| Opus 5 | high | 0.7450 | $0.5352 | 27.7s |
| Sonnet 5 | off | 0.6630 | $0.1694 | 21.6s |
| Sonnet 5 | low | 0.5710 | $0.0850 | 7.9s |
| Sonnet 5 | medium | 0.6050 | $0.1219 | 14.5s |
| Sonnet 5 | high | 0.6620 | $0.1919 | 32.8s |
| Haiku 4.5 | off | 0.4320 | $0.0379 | 9s |
| Haiku 4.5 | low | 0.4950 | $0.0516 | 18.5s |
| Haiku 4.5 | medium | 0.4770 | $0.0618 | 29.5s |
| Haiku 4.5 | high | 0.5010 | $0.0624 | 33.2s |
| GLM | high | 0.6420 | $0.0518 | 30.9s |
| GLM | xhigh | 0.6560 | $0.0910 | 95.6s |
| Qwen | low | 0.6280 | $0.0780 | 32.9s |
| Qwen | high | 0.6910 | $0.1208 | 69.2s |
| Qwen | xhigh | 0.6710 | $0.1143 | 68.6s |
| Kimi K3 | low | 0.5970 | $0.0582 | 9.6s |
| Kimi K3 | high | 0.6460 | $0.0951 | 21.5s |
| Kimi K3 | max | 0.6400 | $0.1179 | 33.9s |
| DeepSeek | off | 0.5810 | $0.0035 | 39.3s |
| DeepSeek | high | 0.6390 | $0.0042 | 89.5s |
| DeepSeek | max | 0.6800 | $0.0075 | 118s |

[^2]: Toast 1 is part of a growing body of work on specialised search agents, alongside [SID-1](https://www.sid.ai/research/sid-1) and Chroma's [Context-1](https://www.trychroma.com/research/context-1). While each takes a different approach, they share the goal of bringing frontier-level retrieval to production at lower cost and latency.

On a variety of deep search benchmarks, it reaches frontier model performance, standing in the same league as GPT-5.6 Sol and comfortably outperforming models such as Kimi K3 or GLM-5.2.

It remains lightweight in doing so. A standard Toast 1 run costs approximately $0.016 - $0.023 per query and has an eight-second median latency. Our highest-quality fusion configuration costs approximately $0.05 - $0.07 per query and has an eleven-second median latency. In practice, among the systems in our evaluation that reached similar performance, Toast 1 was 7–11× cheaper and considerably faster: Frontier-model retrieval agents took between 20 seconds and four minutes on the same evaluation.

## Availability and Pricing

Toast 1 is available immediately through the Mixedbread API at the [discounted launch pricing](https://mixedbread.com/pricing#pricing-rates):
- $0.30 per million input tokens
- $0.036 per million cached input tokens (cache writes are free)
- $0.72 per million output tokens

Mixedbread search invoked by Toast 1 is [priced at a special rate](https://mixedbread.com/pricing#toast-1-search-rate).

### With Your Existing Retrieval Stack

Toast 1 was co-designed with Mixedbread Search's primitives and will be at its strongest performance with it. But we put special care in ensuring that it remains backend agnostic: it can run over your existing retrieval indexes, and does not require migrating your existing backend. We conducted thorough testing to ensure that Toast 1 remains competitive with the performance of frontier models in similar conditions at a fraction of the cost and latency, no matter the provided index.

You can use Toast 1 with our [Chat Completions API](https://mixedbread.com/docs/agent/chat-completions) and add it as a retrieval tool to your existing agentic workflows in just a few minutes. Here is a [golden harness you can use directly](https://github.com/mixedbread-ai/toast-harness).

### With Coding Agents

Let your coding agents handle the integration with `npx skills add mixedbread-ai/skills`. Or use Toast 1 directly as a subagent with our [OpenCode integration](https://mixedbread.com/docs/agent/integrations/opencode).

### With Your Mixedbread Stores

**Python:**
```python
from mixedbread import Mixedbread

    client = Mixedbread()

    results = client.stores.search(
        store_identifiers=["legal-documents"],
        query="does the MSA allow assignment on a change of control?",
        search_options={
            "agentic": True,  # enable Toast 1
        },
    )
```

**TypeScript:**
```typescript
import { Mixedbread } from "@mixedbread/sdk";

    const client = new Mixedbread();

    const results = await client.stores.search({
      store_identifiers: ["legal-documents"],
      query: "does the MSA allow assignment on a change of control?",
      search_options: {
        agentic: true, // enable Toast 1
      },
    });
```

[Get an API key](https://platform.mixedbread.com/platform?next=api-keys) with $5 in included credits to try it out.