The median turn latency is 83.8 seconds per prompt, not per whole build, which derives to 12.6 min across the full nine prompt specification. That is a tenth percentile of 54.4 seconds and a ninetieth percentile of 149.9 seconds across 45 recorded executions.

How reliable is Dyad?

55.6% of recorded executions met the published pass criterion with no operator intervention. 10 were partial and 10 failed outright. Every one of those rows is published on this page.

Can I export the code from Dyad?

There is nothing to export, it is already yours. The code ownership subscore is 98.0 out of 100.

Full-stack app builders / rank 12 of 20 / cycle 04

Dyad benchmark

Runs on your own machine against your own model key, which makes it one of the cheapest routes through the spec and gives it a code ownership score that hosted products cannot match: there is no platform to leave. The trade is variance, since throughput depends on the key you point it at, and there is no managed runtime to catch a bad deploy.

Last run: 15 Aug 2026 09:11 UTCNext scheduled run: n/aOfficial site Dyad reviews and ratings

Composite index

67.49

of 100 | withheld vs. cycle 03

Delta not shown for cycle 04: the roster was re-based from one merged ranking into two class boards, so cycle 03 and cycle 04 ranks are not on the same scale. Resumes cycle 05.

At a glance

Composite index

67.49

rank 12 of 20

Rank

#12

of 20 on the full-stack app builders board

Median turn latency

83.8 s per prompt

12.6 min across the full nine prompt spec, derived

Reliability

55.6%

strict pass rate

SEO score

72.0

0 to 100, SEO and GEO axis

Three strongest axes

Code ownership98.0
Value90.0
Turn latency score77.8

Three weakest axes

Reliability55.6
API and MCP60.0
Integrations62.0

Median turn latency

83.8 s per prompt

p10 54.4 s, p90 149.9 s

Strict pass rate

55.6%

25 pass, 10 partial, 10 fail

Executions recorded

5 runs of 9 prompts

Entry price

No licence fee, bring your own model key

Vendor's own unit and billing period. Never a currency conversion

Per axis subscores

Composite recomputed from these 10 rows: 67.49. That is the arithmetic behind the index, and you can repeat it by hand. Rounding rule: half up at two decimals, applied uniformly to all 37 published products.

Weight table

Per axis subscores against the roster median, and weighted contribution
Axis	Score	Median	Weight %	Contribution
Agent performance	68.0	71.5	18	12.24
Reliability	55.6	71.1	18	10.01
Scalability	64.0	66.0	13	8.32
SEO and GEO	72.0	63.5	12	8.64
API and MCP	60.0	58.0	9	5.40
Integrations	62.0	65.0	8	4.96
Design output	72.0	78.5	7	5.04
Turn latency score	77.8	66.5	7	5.45
Value	90.0	74.0	5	4.50
Code ownership	98.0	73.0	3	2.94

Turn latency distribution

Every recorded prompt duration for Dyad, in seconds. Median wall clock time for one prompt, not for a whole build. The cyan tick on the whisker is the median.

min 18.0 smax 265.4 s

Spread against the cohort

p10 54.4 s / median 83.8 s / p90 149.9 s

Scale is shared with every product in the cohort, so the bars are directly comparable between pages. Ratio of p90 to median is 1.79x.

Outcome split

25 pass / 10 partial / 10 fail

SEO and GEO measurements

Taken from the public marketing page this product generated in prompt 8, fetched with JavaScript disabled.

SEO and GEO leaderboard

HTML bytes, no JS

20.0 KB

20,480 bytes of real text content with scripts blocked

LCP

1,880 ms

Throttled connection, cold cache

CLS

0.040

Cumulative layout shift on load

SEO and GEO subscore

72.0

0 to 100, worth 12 index points on this board

Raw run data: 45 executions

One row per prompt execution. Nothing is aggregated away, nothing is hidden, including the failures.

JSON

Raw per run measurements
Run	Prompt	Prompt target	Started	Duration (s)	Outcome	Note
R01	P01	Tenant and user model	11 Aug 2026 08:00 UTC	143.9	partial	Error envelope inconsistent between two of the endpoints.
R01	P02	Authentication and session	11 Aug 2026 08:04 UTC	51.9	fail	Agent looped on the same edit for the full timeout window of 20 minutes.
R01	P03	Projects CRUD with tenancy	11 Aug 2026 08:07 UTC	129.4	partial	Required one clarifying instruction before the criterion was met.
R01	P04	Public REST API with keys	11 Aug 2026 08:11 UTC	153.9	partial	Pagination off by one at the final page boundary.
R01	P05	Outbound webhooks with signatures	11 Aug 2026 08:15 UTC	79.2	pass
R01	P06	Background jobs	11 Aug 2026 08:19 UTC	77.9	pass
R01	P07	Admin panel	11 Aug 2026 08:22 UTC	78.0	pass
R01	P08	Public marketing surface	11 Aug 2026 08:25 UTC	91.5	partial	Validation added on the write path only. Read path left unguarded.
R01	P09	Hardening pass	11 Aug 2026 08:29 UTC	83.8	pass
R02	P01	Tenant and user model	12 Aug 2026 08:00 UTC	86.5	pass
R02	P02	Authentication and session	12 Aug 2026 08:03 UTC	58.5	fail	Task abandoned by the agent with no artifact produced.
R02	P03	Projects CRUD with tenancy	12 Aug 2026 08:06 UTC	67.4	pass
R02	P04	Public REST API with keys	12 Aug 2026 08:09 UTC	98.3	pass
R02	P05	Outbound webhooks with signatures	12 Aug 2026 08:13 UTC	159.4	pass
R02	P06	Background jobs	12 Aug 2026 08:17 UTC	93.6	fail	Deployed URL returned 500 on every request.
R02	P07	Admin panel	12 Aug 2026 08:21 UTC	78.9	pass
R02	P08	Public marketing surface	12 Aug 2026 08:24 UTC	18.0	pass
R02	P09	Hardening pass	12 Aug 2026 08:27 UTC	89.6	fail	Generated migration conflicted with the existing schema and was not reconciled.
R03	P01	Tenant and user model	13 Aug 2026 08:00 UTC	53.4	partial	Error envelope inconsistent between two of the endpoints.
R03	P02	Authentication and session	13 Aug 2026 08:02 UTC	87.7	pass
R03	P03	Projects CRUD with tenancy	13 Aug 2026 08:06 UTC	184.7	partial	Validation added on the write path only. Read path left unguarded.
R03	P04	Public REST API with keys	13 Aug 2026 08:11 UTC	100.4	fail	Deployed URL returned 500 on every request.
R03	P05	Outbound webhooks with signatures	13 Aug 2026 08:15 UTC	265.4	partial	Pagination off by one at the final page boundary.
R03	P06	Background jobs	13 Aug 2026 08:21 UTC	81.3	pass
R03	P07	Admin panel	13 Aug 2026 08:24 UTC	129.1	pass
R03	P08	Public marketing surface	13 Aug 2026 08:29 UTC	30.9	fail	Generated migration conflicted with the existing schema and was not reconciled.
R03	P09	Hardening pass	13 Aug 2026 08:31 UTC	78.3	pass
R04	P01	Tenant and user model	14 Aug 2026 08:00 UTC	80.8	pass
R04	P02	Authentication and session	14 Aug 2026 08:03 UTC	70.7	pass
R04	P03	Projects CRUD with tenancy	14 Aug 2026 08:06 UTC	56.0	partial	Error envelope inconsistent between two of the endpoints.
R04	P04	Public REST API with keys	14 Aug 2026 08:09 UTC	138.3	pass
R04	P05	Outbound webhooks with signatures	14 Aug 2026 08:13 UTC	59.2	pass
R04	P06	Background jobs	14 Aug 2026 08:16 UTC	124.8	fail	Agent looped on the same edit for the full timeout window of 20 minutes.
R04	P07	Admin panel	14 Aug 2026 08:20 UTC	64.2	pass
R04	P08	Public marketing surface	14 Aug 2026 08:23 UTC	95.5	pass
R04	P09	Hardening pass	14 Aug 2026 08:27 UTC	110.4	pass
R05	P01	Tenant and user model	15 Aug 2026 08:00 UTC	97.2	pass
R05	P02	Authentication and session	15 Aug 2026 08:03 UTC	41.4	pass
R05	P03	Projects CRUD with tenancy	15 Aug 2026 08:06 UTC	95.1	partial	Validation added on the write path only. Read path left unguarded.
R05	P04	Public REST API with keys	15 Aug 2026 08:09 UTC	206.7	partial	Error envelope inconsistent between two of the endpoints.
R05	P05	Outbound webhooks with signatures	15 Aug 2026 08:15 UTC	61.7	fail	Build did not start after deploy. Recorded as a fail with the log preserved.
R05	P06	Background jobs	15 Aug 2026 08:18 UTC	108.3	pass
R05	P07	Admin panel	15 Aug 2026 08:22 UTC	62.8	fail	Deployed URL returned 500 on every request.
R05	P08	Public marketing surface	15 Aug 2026 08:25 UTC	58.6	pass
R05	P09	Hardening pass	15 Aug 2026 08:28 UTC	71.5	fail	Build did not start after deploy. Recorded as a fail with the log preserved.

Pricing

Free and open source with your own model key, so the only cost is the tokens you spend. A hosted Pro tier is also available from USD 20 per month for a managed model key and extra features.

Entry price: No licence fee, bring your own model key
Priced from: vendor pricing page
Value subscore: 90.0
Index points per euro: No licence fee

Entry price is the vendor's own published figure, fetched 21 Aug 2026, never a currency conversion. The currency label is the vendor's own except where the pricing note above records that this lab inferred it. Index points per euro is computed from a separate, euro-normalised figure used only for the Value axis: Converted at the European Central Bank euro reference rate of 1.1699 US dollars to the euro, August 21, 2026.

Capabilities as measured

stack: Local project, framework of your choice
deploy: You deploy it, no hosting included
database: Bring your own
auth: Bring your own
background jobs: Whatever you install
webhooks: Hand written
rest api: Hand written
mcp server: Yes
export code: There is nothing to export, it is already yours
custom domain: Your hosting decides
ssr: Your framework decides

Screenshot

Dyad landing page, captured by the lab on 14 Aug 2026 at 1440 by 900. Stored locally, never hotlinked.

Dyad landing page, captured by the lab at 1440 by 900

Dyad, captured 14 Aug 2026.

Sources and fetch dates

Every claim on this page traces to one of these.

Sources with fetch dates
Source	URL	Fetched
Dyad pricing page	https://dyad.sh/pricing	14 Aug 2026
Dyad product site	https://dyad.sh	14 Aug 2026
LLM Tier run harness output	https://www.llm-tier.com/api/runs.json	14 Aug 2026

Rich results carry this composite as 3.4 on a 0 to 5 scale because that is the only scale the format accepts. The number on this page, 0 to 100, is the real one.

Dyad head to head

dyad vs bolt-diy

Reader index

Voted by readers with an account, one score per person per axis. Published beside the lab index, never inside it.

Score it or read the notes

Reader index

72.53

Lab index

67.49

3 readers, 28 axis scores, last change 29 Aug 2026. Readers rate Dyad 5.0 points above the measurement.

Seeded at launch

The reader scores, lab notes and open ratings published here were written by this lab in August 2026 to populate the reader column before the site had readers. They are not customer submissions, they are not vendor supplied and nobody was paid for them. Anything submitted since then comes from a registered account. The lab index is measured from logged runs and is unaffected either way.

Reader mean against the measured lab subscore for each axis
Axis	Weight	Lab	Readers	Votes
Agent performance	18%	68.0	78.3	3
Reliability	18%	55.6	62.7	3
Scalability	13%	64.0	68.0	1
SEO and GEO	12%	72.0	79.3	3
API and MCP	9%	60.0	66.3	3
Integrations	8%	62.0	74.7	3
Design output	7%	72.0	80.0	3
Turn latency score	7%	77.8	46.3	3
Value	5%	90.0	100.0	3
Code ownership	3%	98.0	100.0	3

The reader index is a weighted mean of reader submitted axis scores using the published index weights, renormalised over the axes readers have scored. It is never blended into the lab index. How this works.

Open rating mean

n/a

0 published open ratings, submitted without an account

Lab index

67.49

Measured, not voted

Gap to the measurement

n/a

Positive means open raters put it above the measurement

What agent performance looked like in practice

Across 45 recorded executions this product met the published pass criterion cleanly 25 times, partially 10 times and failed 10 times. The rows above carry the reviewer note for every non clean outcome, which is the only way to tell a product that fails loudly from one that fails quietly.