Introducing

Autoroute KiCad boards in minutes, not days.

Connect your agent

Sign in and create an API key on your account page, then save it as $HOME/zap-key.txt and run chmod 600 $HOME/zap-key.txt.

No key yet? Any MCP client: npx -y zap-pcb-mcp routes small boards (up to 300 pads, 2 or 4 copper layers) through the free trial, no sign-up.

terminal, then restart Claude Code

mkdir -p $HOME/zap && cd $HOME/zap && curl -sO https://zap.mutual.solutions/downloads/zap_mcp.py
claude mcp add zap -e ZAP_API_KEY_FILE=$HOME/zap-key.txt -- python3 $HOME/zap/zap_mcp.py

$HOME/.cursor/mcp.json, then restart Cursor

{"mcpServers": {"zap": {"command": "python3",
  "args": ["/absolute/path/to/zap_mcp.py"],
  "env": {"ZAP_API_KEY_FILE": "/absolute/path/to/zap-key.txt"}}}}

$HOME/.codex/config.toml, then restart Codex

[mcp_servers.zap]
command = "python3"
args = ["/absolute/path/to/zap_mcp.py"]
env = { ZAP_API_KEY_FILE = "/absolute/path/to/zap-key.txt" }

terminal (Claude Code), then /mcp to sign in

claude mcp add --transport http zap https://zap.mutual.solutions/mcp

terminal

curl -sO https://zap.mutual.solutions/downloads/zap.py
export ZAP_API_KEY_FILE=$HOME/zap-key.txt
python3 zap.py run myboard.kicad_pcb

Then restart the agent.

Upload a board All 25 boards

FTDI JTAG programmer

Lower is better

1 CPU core each, up to 60 minutes per board.

Zap!

Time: 20 s
Unrouted: 0
Errors: 1

TraceMaker

Time: 72 s
Unrouted: 0
Errors: 3

tscircuit

Time: 89 s
Unrouted: 0
Errors: 419

Freerouting

Time: 205 s
Unrouted: 11
Errors: 4

FTDI JTAG programmer by Anthony Bernaert · CERN-OHL-P-2.0 · source

Your rules, in full

Rule types it cannot take · lower is better

Zap!0

TraceMaker1

Freerouting9

tscircuit18

	Zap!	TraceMaker	Freerouting	tscircuit
Track and via spacing	full	full	approximated	cannot be given
Net-class widths and via sizes	full	full	full	approximated
Differential pairs: coupling and length	full	full	cannot be given	approximated
Copper pours	full	cannot be given	cannot be given	cannot be given
Keep-out areas	full	full	approximated	approximated
Custom rules (.kicad_dru)	full	full	cannot be given	cannot be given

fullapproximatedcannot be given

All rule types

tscircuit: 18 rule types

rule type	boards
Rounded, oval and chamfered pads are given as rectangles of the pad size, corners filled in (the input format has no rounded shape; tscircuit's own KiCad import does the same)	25
No track-to-track or via-to-via spacing rule; the pad spacing is used instead, and some solver stages use a fixed 0.1 mm	25
No copper pours: power and ground are routed as tracks	25
One track width per connection: no width ranges, neck-downs or length tuning	25
No length-matching limits for differential pairs	17
Differential pairs that reach more than two pads are routed as ordinary nets	12
Keep-out areas that only forbid pours or parts are ignored (it makes no pours)	12
Custom-shape pads are covered by rectangles that follow the pad outline (exact where its edges are straight and square)	7
Spacing overrides on single pads or parts are dropped	5
Odd-shaped copper inside parts is covered by rectangles that follow its outline	5
One spacing for every net class (the default class)	3
The board's custom design rules (.kicad_dru) cannot be expressed	2
Odd-shaped keep-out areas are covered by rectangles that follow their outline	2
Internal cut-outs are blocked on every layer	2
Via-only keep-outs are left out (vias placed there count as errors)	2
One via size for every net class (the default class)	1
Track-only keep-outs also block vias (stricter than the board)	1
Odd-shaped board cut-outs are covered by rectangles that follow their outline	1

Freerouting: 9 rule types

rule type	boards
No hole clearance rule (KiCad's minimum hole clearance; Freerouting used 0)	25
No minimum hole-to-hole distance	25
No copper pours: power and ground are routed as tracks	25
No length or skew rules (net-class or custom)	25
Differential pairs: no coupling, gap, uncoupled-length or skew rule; routed as independent nets	17
Keep-out areas that only forbid pours or parts were removed so tracks may pass, as KiCad allows	11
Copper-to-edge spacing above 0.25 mm not passed on (Freerouting used 0.25 mm)	10
The board's custom design rules (.kicad_dru) are not exported	2
Track-only keep-outs also block vias (stricter than the board)	1

TraceMaker: 1 rule type

rule type	boards
No copper pours: power and ground are routed as tracks	25

TraceMaker reported no other board rule it could not evaluate on the 25 public boards; it writes no copper pours.

Zap! reads KiCad net classes and custom rules itself and refuses a job whose rules it cannot read; it refused none of the 25 boards.

Every measure

Unrouted connections

total over 25 boards · lower is better

Zap! 60 minutes124

Connections routed on each board

largest board first

routedhigher is better

Unrouted connections on each board

largest board first

countlower is better

Shorts on each board

largest board first

countlower is better

Design-rule errors on each board

largest board first

countlower is better

Seconds on each board

largest board first

secondslower is better

Antmicro ECP5 DC-SCM93.8%11305093780.9%352023timed outgave upgave upgave upgave upgave up1.3%1,81808timed out100%009–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	113	0	50	937
TraceMaker	352	0	23	timed out
tscircuit	1,843	gave up	gave up	gave up
Freerouting	1,818	0	8	timed out
Human designs	0	0	9	–

tscircuit routed no connection. Error: its pre-check reported a connection as impossible, but the designers' own route for it fits the converted board.

Antmicro ECP5 DC-SCM as its designers routed it, all copper layers — Human designs

Antmicro ECP5 DC-SCM as its designers routed it, F.Cu — Human designs

Antmicro ECP5 DC-SCM routed by Zap!, all copper layers — Zap!

Antmicro ECP5 DC-SCM routed by Zap!, F.Cu — Zap!

Antmicro ECP5 DC-SCM routed by TraceMaker, all copper layers — TraceMaker

Antmicro ECP5 DC-SCM routed by TraceMaker, F.Cu — TraceMaker

Antmicro ECP5 DC-SCM routed by tscircuit, all copper layers — tscircuit

Antmicro ECP5 DC-SCM routed by tscircuit, F.Cu — tscircuit

Antmicro ECP5 DC-SCM routed by Freerouting, all copper layers — Freerouting

Antmicro ECP5 DC-SCM routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Antmicro ECP5 DC-SCM by Antmicro · Apache-2.0 · source

HackRF Pro99.3%110964292.2%13009timed outgave upgave upgave upgave upgave up95.2%800210timed out100%008–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	11	0	9	642
TraceMaker	130	0	9	timed out
tscircuit	1,364	gave up	gave up	gave up
Freerouting	80	0	210	timed out
Human designs	0	0	8	–

tscircuit routed no connection. Error: its pre-check reported a connection as impossible, but the designers' own route for it fits the converted board.

HackRF Pro as its designers routed it, all copper layers — Human designs

HackRF Pro as its designers routed it, F.Cu — Human designs

HackRF Pro routed by Zap!, all copper layers — Zap!

HackRF Pro routed by TraceMaker, all copper layers — TraceMaker

HackRF Pro routed by TraceMaker, F.Cu — TraceMaker

HackRF Pro routed by tscircuit, all copper layers — tscircuit

HackRF Pro routed by tscircuit, F.Cu — tscircuit

HackRF Pro routed by Freerouting, all copper layers — Freerouting

HackRF Pro routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

HackRF Pro by Great Scott Gadgets · CERN-OHL-P-2.0 · source

EmberOne99.5%50144598.4%1800timed out98.2%2059799577100%002182,653100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	5	0	1	445
TraceMaker	18	0	0	timed out
tscircuit	20	59	799	577
Freerouting	0	0	218	2,653
Human designs	0	0	0	–

Numbers only: its licence (CERN-OHL-S) asks that modified designs be shared in full. Source.

Buildbotics controller99.2%70451594.8%4604timed out99.8%12178991,38397.0%260203timed out100%005–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	7	0	4	515
TraceMaker	46	0	4	timed out
tscircuit	1	217	899	1,383
Freerouting	26	0	203	timed out
Human designs	0	0	5	–

Numbers only: its licence (CERN-OHL-S) asks that modified designs be shared in full. Source.

Kintex PCIe board99.1%603056791.9%57055timed outtimed outtimed outtimed outtimed outtimed out100%01263timed out100%0018–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	6	0	30	567
TraceMaker	57	0	55	timed out
tscircuit	706	timed out	timed out	timed out
Freerouting	0	1	263	timed out
Human designs	0	0	18	–

tscircuit routed no connection. Killed at the 75-minute backstop, past its 60-minute time limit, while still inside one solver step; no routing written.

Kintex PCIe board as its designers routed it, all copper layers — Human designs

Kintex PCIe board as its designers routed it, F.Cu — Human designs

Kintex PCIe board routed by Zap!, all copper layers — Zap!

Kintex PCIe board routed by Zap!, F.Cu — Zap!

Kintex PCIe board routed by TraceMaker, all copper layers — TraceMaker

Kintex PCIe board routed by TraceMaker, F.Cu — TraceMaker

Kintex PCIe board routed by tscircuit, all copper layers — tscircuit

Kintex PCIe board routed by tscircuit, F.Cu — tscircuit

Kintex PCIe board routed by Freerouting, all copper layers — Freerouting

Kintex PCIe board routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Kintex PCIe board by Piotr Binkowski · Apache-2.0 · source

discretesim test board100%0004095.7%25003,00799.6%21656613989.3%6200613100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	0	40
TraceMaker	25	0	0	3,007
tscircuit	2	16	566	139
Freerouting	62	0	0	613
Human designs	0	0	0	–

discretesim test board as its designers routed it, all copper layers — Human designs

discretesim test board as its designers routed it, F.Cu — Human designs

discretesim test board routed by Zap!, all copper layers — Zap!

discretesim test board routed by Zap!, F.Cu — Zap!

discretesim test board routed by TraceMaker, all copper layers — TraceMaker

discretesim test board routed by TraceMaker, F.Cu — TraceMaker

discretesim test board routed by tscircuit, all copper layers — tscircuit

discretesim test board routed by tscircuit, F.Cu — tscircuit

discretesim test board routed by Freerouting, all copper layers — Freerouting

discretesim test board routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

discretesim test board by cnlohr · MIT · source

Antmicro STM32H7 HDMI SO-DIMM board100%00421993.6%360203timed out87.4%712011,122470100%00240timed out100%004–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	4	219
TraceMaker	36	0	203	timed out
tscircuit	71	201	1,122	470
Freerouting	0	0	240	timed out
Human designs	0	0	4	–

Antmicro STM32H7 HDMI SO-DIMM board as its designers routed it, all copper layers — Human designs

Antmicro STM32H7 HDMI SO-DIMM board as its designers routed it, F.Cu — Human designs

Antmicro STM32H7 HDMI SO-DIMM board routed by Zap!, all copper layers — Zap!

Antmicro STM32H7 HDMI SO-DIMM board routed by Zap!, F.Cu — Zap!

Antmicro STM32H7 HDMI SO-DIMM board routed by TraceMaker, all copper layers — TraceMaker

Antmicro STM32H7 HDMI SO-DIMM board routed by TraceMaker, F.Cu — TraceMaker

Antmicro STM32H7 HDMI SO-DIMM board routed by tscircuit, all copper layers — tscircuit

Antmicro STM32H7 HDMI SO-DIMM board routed by tscircuit, F.Cu — tscircuit

Antmicro STM32H7 HDMI SO-DIMM board routed by Freerouting, all copper layers — Freerouting

Antmicro STM32H7 HDMI SO-DIMM board routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Antmicro STM32H7 HDMI SO-DIMM board by Antmicro · Apache-2.0 · source

Tiliqua (apfaudio)99.6%201753392.0%42016timed outcrashedcrashedcrashedcrashedcrashed100%002421,727100%0014–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	2	0	17	533
TraceMaker	42	0	16	timed out
tscircuit	510	crashed	crashed	crashed
Freerouting	0	0	242	1,727
Human designs	0	0	14	–

tscircuit routed no connection. Error after 1442 s: repair04 merge rejected a new via in the boundary collar.

Numbers only: its licence (CERN-OHL-S) asks that modified designs be shared in full. Source.

Antmicro CM4 baseboard LVDS adapter100%001347899.7%109timed out99.7%1338943998.4%60137733100%001–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	13	478
TraceMaker	1	0	9	timed out
tscircuit	1	3	389	439
Freerouting	6	0	137	733
Human designs	0	0	1	–

Antmicro CM4 baseboard LVDS adapter as its designers routed it, all copper layers — Human designs

Antmicro CM4 baseboard LVDS adapter as its designers routed it, F.Cu — Human designs

Antmicro CM4 baseboard LVDS adapter routed by Zap!, all copper layers — Zap!

Antmicro CM4 baseboard LVDS adapter routed by Zap!, F.Cu — Zap!

Antmicro CM4 baseboard LVDS adapter routed by TraceMaker, all copper layers — TraceMaker

Antmicro CM4 baseboard LVDS adapter routed by TraceMaker, F.Cu — TraceMaker

Antmicro CM4 baseboard LVDS adapter routed by tscircuit, all copper layers — tscircuit

Antmicro CM4 baseboard LVDS adapter routed by tscircuit, F.Cu — tscircuit

Antmicro CM4 baseboard LVDS adapter routed by Freerouting, all copper layers — Freerouting

Antmicro CM4 baseboard LVDS adapter routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Antmicro CM4 baseboard LVDS adapter by Antmicro · Apache-2.0 · source

Bitaxe Ultra (fork)100%00342099.0%303timed out100%02444425699.6%116986100%003–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	3	420
TraceMaker	3	0	3	timed out
tscircuit	0	24	444	256
Freerouting	1	1	69	86
Human designs	0	0	3	–

Numbers only: its licence (CERN-OHL-S) asks that modified designs be shared in full. Source.

Xerxes96.4%1101551891.4%27013timed out92.0%2518481234100%00237172100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	11	0	15	518
TraceMaker	27	0	13	timed out
tscircuit	25	18	481	234
Freerouting	0	0	237	172
Human designs	0	0	0	–

Xerxes as its designers routed it, all copper layers — Human designs

Xerxes as its designers routed it, F.Cu — Human designs

Xerxes routed by Zap!, all copper layers — Zap!

Xerxes routed by TraceMaker, all copper layers — TraceMaker

Xerxes routed by TraceMaker, F.Cu — TraceMaker

Xerxes routed by tscircuit, all copper layers — tscircuit

Xerxes routed by tscircuit, F.Cu — tscircuit

Xerxes routed by Freerouting, all copper layers — Freerouting

Xerxes routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Xerxes by Rapid Analysis · Apache-2.0 · source

RP2350-PICO2-BB4887.3%340237688.8%3002timed out98.8%32151,02894197.0%80245557100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	34	0	2	376
TraceMaker	30	0	2	timed out
tscircuit	3	215	1,028	941
Freerouting	8	0	245	557
Human designs	0	0	0	–

RP2350-PICO2-BB48 as its designers routed it, all copper layers — Human designs

RP2350-PICO2-BB48 as its designers routed it, F.Cu — Human designs

RP2350-PICO2-BB48 routed by Zap!, all copper layers — Zap!

RP2350-PICO2-BB48 routed by Zap!, F.Cu — Zap!

RP2350-PICO2-BB48 routed by TraceMaker, all copper layers — TraceMaker

RP2350-PICO2-BB48 routed by TraceMaker, F.Cu — TraceMaker

RP2350-PICO2-BB48 routed by tscircuit, all copper layers — tscircuit

RP2350-PICO2-BB48 routed by tscircuit, F.Cu — tscircuit

RP2350-PICO2-BB48 routed by Freerouting, all copper layers — Freerouting

RP2350-PICO2-BB48 routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

RP2350-PICO2-BB48 by Olimex Ltd · MIT · source

FTDI JTAG programmer100%00120100%00372100%0214198995.8%1104205100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	1	20
TraceMaker	0	0	3	72
tscircuit	0	21	419	89
Freerouting	11	0	4	205
Human designs	0	0	0	–

FTDI JTAG programmer as its designers routed it, all copper layers — Human designs

FTDI JTAG programmer as its designers routed it, F.Cu — Human designs

FTDI JTAG programmer routed by Zap!, all copper layers — Zap!

FTDI JTAG programmer routed by Zap!, F.Cu — Zap!

FTDI JTAG programmer routed by TraceMaker, all copper layers — TraceMaker

FTDI JTAG programmer routed by TraceMaker, F.Cu — TraceMaker

FTDI JTAG programmer routed by tscircuit, all copper layers — tscircuit

FTDI JTAG programmer routed by tscircuit, F.Cu — tscircuit

FTDI JTAG programmer routed by Freerouting, all copper layers — Freerouting

FTDI JTAG programmer routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

FTDI JTAG programmer by Anthony Bernaert · CERN-OHL-P-2.0 · source

Antmicro GMSL serializer98.4%401413199.2%20133,382step limitstep limitstep limitstep limitstep limit92.1%20018547100%007–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	4	0	14	131
TraceMaker	2	0	13	3,382
tscircuit	252	step limit	step limit	step limit
Freerouting	20	0	18	547
Human designs	0	0	7	–

tscircuit routed no connection. Limit: a fixed step limit inside one solver stage (it ran out of iterations).

Antmicro GMSL serializer as its designers routed it, all copper layers — Human designs

Antmicro GMSL serializer as its designers routed it, F.Cu — Human designs

Antmicro GMSL serializer routed by Zap!, all copper layers — Zap!

Antmicro GMSL serializer routed by Zap!, F.Cu — Zap!

Antmicro GMSL serializer routed by TraceMaker, all copper layers — TraceMaker

Antmicro GMSL serializer routed by TraceMaker, F.Cu — TraceMaker

Antmicro GMSL serializer routed by tscircuit, all copper layers — tscircuit

Antmicro GMSL serializer routed by tscircuit, F.Cu — tscircuit

Antmicro GMSL serializer routed by Freerouting, all copper layers — Freerouting

Antmicro GMSL serializer routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Antmicro GMSL serializer by Antmicro · Apache-2.0 · source

Seven99.1%20040294.1%1402timed out100%02554013092.5%1807692100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	2	0	0	402
TraceMaker	14	0	2	timed out
tscircuit	0	25	540	130
Freerouting	18	0	7	692
Human designs	0	0	0	–

Seven as its designers routed it, all copper layers — Human designs

Seven as its designers routed it, F.Cu — Human designs

Seven routed by Zap!, all copper layers — Zap!

Seven routed by TraceMaker, all copper layers — TraceMaker

Seven routed by TraceMaker, F.Cu — TraceMaker

Seven routed by tscircuit, all copper layers — tscircuit

Seven routed by tscircuit, F.Cu — tscircuit

Seven routed by Freerouting, all copper layers — Freerouting

Seven routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Seven by ID8 Engineering AB · CERN-OHL-P-2.0 · source

Argos demo board100%00024100%000511100%018114104100%001104100%000–

router	Shorts	Rule errors	Seconds
Zap!	0	0	24
TraceMaker	0	0	511
tscircuit	18	114	104
Freerouting	0	1	104
Human designs	0	0	–

Numbers only: its licence (CERN-OHL-S) asks that modified designs be shared in full. Source.

Antmicro CVBS-to-MIPI bridge100%00210998.5%301timed out100%03224014498.1%4016410100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	2	109
TraceMaker	3	0	1	timed out
tscircuit	0	32	240	144
Freerouting	4	0	16	410
Human designs	0	0	0	–

Antmicro CVBS-to-MIPI bridge as its designers routed it, all copper layers — Human designs

Antmicro CVBS-to-MIPI bridge as its designers routed it, F.Cu — Human designs

Antmicro CVBS-to-MIPI bridge routed by Zap!, all copper layers — Zap!

Antmicro CVBS-to-MIPI bridge routed by Zap!, F.Cu — Zap!

Antmicro CVBS-to-MIPI bridge routed by TraceMaker, all copper layers — TraceMaker

Antmicro CVBS-to-MIPI bridge routed by TraceMaker, F.Cu — TraceMaker

Antmicro CVBS-to-MIPI bridge routed by tscircuit, all copper layers — tscircuit

Antmicro CVBS-to-MIPI bridge routed by tscircuit, F.Cu — tscircuit

Antmicro CVBS-to-MIPI bridge routed by Freerouting, all copper layers — Freerouting

Antmicro CVBS-to-MIPI bridge routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Antmicro CVBS-to-MIPI bridge by Antmicro · Apache-2.0 · source

Trellis Core100%00020100%000559100%0213080100%002130100%000–

router	Shorts	Rule errors	Seconds
Zap!	0	0	20
TraceMaker	0	0	559
tscircuit	2	130	80
Freerouting	0	2	130
Human designs	0	0	–

Trellis Core as its designers routed it, all copper layers — Human designs

Trellis Core as its designers routed it, F.Cu — Human designs

Trellis Core routed by Zap!, all copper layers — Zap!

Trellis Core routed by Zap!, F.Cu — Zap!

Trellis Core routed by TraceMaker, all copper layers — TraceMaker

Trellis Core routed by TraceMaker, F.Cu — TraceMaker

Trellis Core routed by tscircuit, all copper layers — tscircuit

Trellis Core routed by tscircuit, F.Cu — tscircuit

Trellis Core routed by Freerouting, all copper layers — Freerouting

Trellis Core routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Trellis Core by Protolux Electronics SARL-S · CERN-OHL-P-2.0 · source

Antmicro PMOD I3C sensor board100%00710100%00732497.2%5382324299.4%1286414100%007–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	7	10
TraceMaker	0	0	7	324
tscircuit	5	38	232	42
Freerouting	1	2	86	414
Human designs	0	0	7	–

Antmicro PMOD I3C sensor board as its designers routed it, all copper layers — Human designs

Antmicro PMOD I3C sensor board as its designers routed it, F.Cu — Human designs

Antmicro PMOD I3C sensor board routed by Zap!, all copper layers — Zap!

Antmicro PMOD I3C sensor board routed by Zap!, F.Cu — Zap!

Antmicro PMOD I3C sensor board routed by TraceMaker, all copper layers — TraceMaker

Antmicro PMOD I3C sensor board routed by TraceMaker, F.Cu — TraceMaker

Antmicro PMOD I3C sensor board routed by tscircuit, all copper layers — tscircuit

Antmicro PMOD I3C sensor board routed by tscircuit, F.Cu — tscircuit

Antmicro PMOD I3C sensor board routed by Freerouting, all copper layers — Freerouting

Antmicro PMOD I3C sensor board routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Antmicro PMOD I3C sensor board by Antmicro · Apache-2.0 · source

Ender-3 load-cell bed-levelling sensor100%00229100%002109100%01328510399.3%10271100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	2	29
TraceMaker	0	0	2	109
tscircuit	0	13	285	103
Freerouting	1	0	2	71
Human designs	0	0	0	–

Ender-3 load-cell bed-levelling sensor as its designers routed it, all copper layers — Human designs

Ender-3 load-cell bed-levelling sensor as its designers routed it, F.Cu — Human designs

Ender-3 load-cell bed-levelling sensor routed by Zap!, all copper layers — Zap!

Ender-3 load-cell bed-levelling sensor routed by Zap!, F.Cu — Zap!

Ender-3 load-cell bed-levelling sensor routed by TraceMaker, all copper layers — TraceMaker

Ender-3 load-cell bed-levelling sensor routed by TraceMaker, F.Cu — TraceMaker

Ender-3 load-cell bed-levelling sensor routed by tscircuit, all copper layers — tscircuit

Ender-3 load-cell bed-levelling sensor routed by tscircuit, F.Cu — tscircuit

Ender-3 load-cell bed-levelling sensor routed by Freerouting, all copper layers — Freerouting

Ender-3 load-cell bed-levelling sensor routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Ender-3 load-cell bed-levelling sensor by Lakshminarayanan · MIT · source

ECP5 BGA-381 FT2232HQ FIFO board100%0006995.8%600timed out82.6%2502663100%000120100%000–

router	Unrouted	Rule errors	Seconds
Zap!	0	0	69
TraceMaker	6	0	timed out
tscircuit	25	26	63
Freerouting	0	0	120
Human designs	0	0	–

ECP5 BGA-381 FT2232HQ FIFO board as its designers routed it, all copper layers — Human designs

ECP5 BGA-381 FT2232HQ FIFO board as its designers routed it, F.Cu — Human designs

ECP5 BGA-381 FT2232HQ FIFO board routed by Zap!, all copper layers — Zap!

ECP5 BGA-381 FT2232HQ FIFO board routed by Zap!, F.Cu — Zap!

ECP5 BGA-381 FT2232HQ FIFO board routed by TraceMaker, all copper layers — TraceMaker

ECP5 BGA-381 FT2232HQ FIFO board routed by TraceMaker, F.Cu — TraceMaker

ECP5 BGA-381 FT2232HQ FIFO board routed by tscircuit, all copper layers — tscircuit

ECP5 BGA-381 FT2232HQ FIFO board routed by tscircuit, F.Cu — tscircuit

ECP5 BGA-381 FT2232HQ FIFO board routed by Freerouting, all copper layers — Freerouting

ECP5 BGA-381 FT2232HQ FIFO board routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

ECP5 BGA-381 FT2232HQ FIFO board by gildobjanschi · MIT · source

RP2040 development board99.2%10418492.0%11052,174100%016279351887.7%17035257100%003–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	1	0	4	184
TraceMaker	11	0	5	2,174
tscircuit	0	162	793	518
Freerouting	17	0	35	257
Human designs	0	0	3	–

Numbers only: by decision. Source.

Antmicro signal-integrity test board100%0008895.3%50031966.3%36172034100%000529100%003–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	0	88
TraceMaker	5	0	0	319
tscircuit	36	17	20	34
Freerouting	0	0	0	529
Human designs	0	0	3	–

Antmicro signal-integrity test board as its designers routed it, all copper layers — Human designs

Antmicro signal-integrity test board as its designers routed it, F.Cu — Human designs

Antmicro signal-integrity test board routed by Zap!, all copper layers — Zap!

Antmicro signal-integrity test board routed by Zap!, F.Cu — Zap!

Antmicro signal-integrity test board routed by TraceMaker, all copper layers — TraceMaker

Antmicro signal-integrity test board routed by TraceMaker, F.Cu — TraceMaker

Antmicro signal-integrity test board routed by tscircuit, all copper layers — tscircuit

Antmicro signal-integrity test board routed by tscircuit, F.Cu — tscircuit

Antmicro signal-integrity test board routed by Freerouting, all copper layers — Freerouting

Antmicro signal-integrity test board routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Antmicro signal-integrity test board by Antmicro · Apache-2.0 · source

Antmicro dual IPEX CSI interposer100%00933100%0069412100%003561100%006871100%000–

router	Rule errors	Seconds
Zap!	9	33
TraceMaker	69	412
tscircuit	35	61
Freerouting	68	71
Human designs	0	–

Antmicro dual IPEX CSI interposer as its designers routed it, all copper layers — Human designs

Antmicro dual IPEX CSI interposer as its designers routed it, F.Cu — Human designs

Antmicro dual IPEX CSI interposer routed by Zap!, all copper layers — Zap!

Antmicro dual IPEX CSI interposer routed by Zap!, F.Cu — Zap!

Antmicro dual IPEX CSI interposer routed by TraceMaker, all copper layers — TraceMaker

Antmicro dual IPEX CSI interposer routed by TraceMaker, F.Cu — TraceMaker

Antmicro dual IPEX CSI interposer routed by tscircuit, all copper layers — tscircuit

Antmicro dual IPEX CSI interposer routed by tscircuit, F.Cu — tscircuit

Antmicro dual IPEX CSI interposer routed by Freerouting, all copper layers — Freerouting

Antmicro dual IPEX CSI interposer routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Antmicro dual IPEX CSI interposer by Antmicro · Apache-2.0 · source

Arduino water-level monitor100%0005100%00011100%0038100%00717100%000–

router	Rule errors	Seconds
Zap!	0	5
TraceMaker	0	11
tscircuit	3	8
Freerouting	7	17
Human designs	0	–

Arduino water-level monitor as its designers routed it, all copper layers — Human designs

Arduino water-level monitor as its designers routed it, F.Cu — Human designs

Arduino water-level monitor routed by Zap!, all copper layers — Zap!

Arduino water-level monitor routed by Zap!, F.Cu — Zap!

Arduino water-level monitor routed by TraceMaker, all copper layers — TraceMaker

Arduino water-level monitor routed by TraceMaker, F.Cu — TraceMaker

Arduino water-level monitor routed by tscircuit, all copper layers — tscircuit

Arduino water-level monitor routed by tscircuit, F.Cu — tscircuit

Arduino water-level monitor routed by Freerouting, all copper layers — Freerouting

Arduino water-level monitor routed by Freerouting, F.Cu — Freerouting

KiCad layer colours · grey: board edge · dark circles: drill holes

Arduino water-level monitor by Srushti D · MIT · source

Unrouted connections

total over 25 boards · lower is better

Connections routed on each board

largest board first

routedhigher is better

Unrouted connections on each board

largest board first

countlower is better

Shorts on each board

largest board first

countlower is better

Design-rule errors on each board

largest board first

countlower is better

Seconds on each board

largest board first

secondslower is better

Antmicro ECP5 DC-SCM96.6%61031timed out87.0%239026timed outgave upgave upgave upgave upgave up1.0%1,82406timed out100%009–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	61	0	31	timed out
TraceMaker	239	0	26	timed out
tscircuit	1,843	gave up	gave up	gave up
Freerouting	1,824	0	6	timed out
Human designs	0	0	9	–

tscircuit routed no connection. Error: its pre-check reported a connection as impossible, but the designers' own route for it fits the converted board.

Antmicro ECP5 DC-SCM as its designers routed it — Human designs

Antmicro ECP5 DC-SCM routed by Zap! — Zap!

Antmicro ECP5 DC-SCM routed by TraceMaker — TraceMaker

Antmicro ECP5 DC-SCM routed by tscircuit — tscircuit

Antmicro ECP5 DC-SCM routed by Freerouting — Freerouting

Antmicro ECP5 DC-SCM by Antmicro · Apache-2.0 · source

HackRF Pro99.7%409timed out94.7%88010timed outgave upgave upgave upgave upgave up93.6%1060208timed out100%008–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	4	0	9	timed out
TraceMaker	88	0	10	timed out
tscircuit	1,364	gave up	gave up	gave up
Freerouting	106	0	208	timed out
Human designs	0	0	8	–

tscircuit routed no connection. Error: its pre-check reported a connection as impossible, but the designers' own route for it fits the converted board.

HackRF Pro as its designers routed it — Human designs

HackRF Pro routed by TraceMaker — TraceMaker

HackRF Pro routed by tscircuit — tscircuit

HackRF Pro routed by Freerouting — Freerouting

HackRF Pro by Great Scott Gadgets · CERN-OHL-P-2.0 · source

EmberOne100%00177499.9%100timed out98.2%2059799577100%002182,432100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	1	774
TraceMaker	1	0	0	timed out
tscircuit	20	59	799	577
Freerouting	0	0	218	2,432
Human designs	0	0	0	–

Numbers only: its licence (CERN-OHL-S) asks that modified designs be shared in full. Source.

Buildbotics controller99.6%304timed out98.5%1304timed out99.8%12178991,38397.1%250203timed out100%005–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	3	0	4	timed out
TraceMaker	13	0	4	timed out
tscircuit	1	217	899	1,383
Freerouting	25	0	203	timed out
Human designs	0	0	5	–

Numbers only: its licence (CERN-OHL-S) asks that modified designs be shared in full. Source.

Kintex PCIe board99.4%4036timed out99.5%3052timed outtimed outtimed outtimed outtimed outtimed out100%012633,116100%0018–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	4	0	36	timed out
TraceMaker	3	0	52	timed out
tscircuit	706	timed out	timed out	timed out
Freerouting	0	1	263	3,116
Human designs	0	0	18	–

tscircuit routed no connection. Killed at the 75-minute backstop, past its 60-minute time limit, while still inside one solver step; no routing written.

Kintex PCIe board as its designers routed it — Human designs

Kintex PCIe board routed by TraceMaker — TraceMaker

Kintex PCIe board routed by tscircuit — tscircuit

Kintex PCIe board routed by Freerouting — Freerouting

Kintex PCIe board by Piotr Binkowski · Apache-2.0 · source

discretesim test board100%0007396.3%21003,25099.6%21656613989.3%6200597100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	0	73
TraceMaker	21	0	0	3,250
tscircuit	2	16	566	139
Freerouting	62	0	0	597
Human designs	0	0	0	–

discretesim test board as its designers routed it — Human designs

discretesim test board routed by Zap! — Zap!

discretesim test board routed by TraceMaker — TraceMaker

discretesim test board routed by tscircuit — tscircuit

discretesim test board routed by Freerouting — Freerouting

discretesim test board by cnlohr · MIT · source

Antmicro STM32H7 HDMI SO-DIMM board100%00437796.6%190203timed out87.4%712011,122470100%002403,290100%004–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	4	377
TraceMaker	19	0	203	timed out
tscircuit	71	201	1,122	470
Freerouting	0	0	240	3,290
Human designs	0	0	4	–

Antmicro STM32H7 HDMI SO-DIMM board as its designers routed it — Human designs

Antmicro STM32H7 HDMI SO-DIMM board routed by Zap! — Zap!

Antmicro STM32H7 HDMI SO-DIMM board routed by TraceMaker — TraceMaker

Antmicro STM32H7 HDMI SO-DIMM board routed by tscircuit — tscircuit

Antmicro STM32H7 HDMI SO-DIMM board routed by Freerouting — Freerouting

Antmicro STM32H7 HDMI SO-DIMM board by Antmicro · Apache-2.0 · source

Tiliqua (apfaudio)100%00181,52299.4%3018timed outcrashedcrashedcrashedcrashedcrashed100%00242476100%0014–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	18	1,522
TraceMaker	3	0	18	timed out
tscircuit	510	crashed	crashed	crashed
Freerouting	0	0	242	476
Human designs	0	0	14	–

tscircuit routed no connection. Error after 1442 s: repair04 merge rejected a new via in the boundary collar.

Numbers only: its licence (CERN-OHL-S) asks that modified designs be shared in full. Source.

Antmicro CM4 baseboard LVDS adapter100%0013790100%001163099.7%1338943998.4%60137552100%001–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	13	790
TraceMaker	0	0	11	630
tscircuit	1	3	389	439
Freerouting	6	0	137	552
Human designs	0	0	1	–

Antmicro CM4 baseboard LVDS adapter as its designers routed it — Human designs

Antmicro CM4 baseboard LVDS adapter routed by Zap! — Zap!

Antmicro CM4 baseboard LVDS adapter routed by TraceMaker — TraceMaker

Antmicro CM4 baseboard LVDS adapter routed by tscircuit — tscircuit

Antmicro CM4 baseboard LVDS adapter routed by Freerouting — Freerouting

Antmicro CM4 baseboard LVDS adapter by Antmicro · Apache-2.0 · source

Bitaxe Ultra (fork)100%00374699.0%303timed out100%02444425699.6%126950100%003–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	3	746
TraceMaker	3	0	3	timed out
tscircuit	0	24	444	256
Freerouting	1	2	69	50
Human designs	0	0	3	–

Numbers only: its licence (CERN-OHL-S) asks that modified designs be shared in full. Source.

Xerxes100%0014timed out100%001462192.0%2518481234100%0023790100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	14	timed out
TraceMaker	0	0	14	621
tscircuit	25	18	481	234
Freerouting	0	0	237	90
Human designs	0	0	0	–

Xerxes as its designers routed it — Human designs

Xerxes routed by TraceMaker — TraceMaker

Xerxes routed by Freerouting — Freerouting

Xerxes by Rapid Analysis · Apache-2.0 · source

RP2350-PICO2-BB4890.7%2503timed out90.7%2502timed out98.8%32151,02894198.5%40254742100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	25	0	3	timed out
TraceMaker	25	0	2	timed out
tscircuit	3	215	1,028	941
Freerouting	4	0	254	742
Human designs	0	0	0	–

RP2350-PICO2-BB48 as its designers routed it — Human designs

RP2350-PICO2-BB48 routed by TraceMaker — TraceMaker

RP2350-PICO2-BB48 routed by tscircuit — tscircuit

RP2350-PICO2-BB48 routed by Freerouting — Freerouting

RP2350-PICO2-BB48 by Olimex Ltd · MIT · source

FTDI JTAG programmer100%00194100%001108100%0214198995.8%1104168100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	1	94
TraceMaker	0	0	1	108
tscircuit	0	21	419	89
Freerouting	11	0	4	168
Human designs	0	0	0	–

FTDI JTAG programmer as its designers routed it — Human designs

FTDI JTAG programmer routed by Zap! — Zap!

FTDI JTAG programmer routed by TraceMaker — TraceMaker

FTDI JTAG programmer routed by tscircuit — tscircuit

FTDI JTAG programmer routed by Freerouting — Freerouting

FTDI JTAG programmer by Anthony Bernaert · CERN-OHL-P-2.0 · source

Antmicro GMSL serializer99.6%1014timed out99.2%2013timed outstep limitstep limitstep limitstep limitstep limit92.1%20018337100%007–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	1	0	14	timed out
TraceMaker	2	0	13	timed out
tscircuit	252	step limit	step limit	step limit
Freerouting	20	0	18	337
Human designs	0	0	7	–

tscircuit routed no connection. Limit: a fixed step limit inside one solver stage (it ran out of iterations).

Antmicro GMSL serializer as its designers routed it — Human designs

Antmicro GMSL serializer routed by Zap! — Zap!

Antmicro GMSL serializer routed by TraceMaker — TraceMaker

Antmicro GMSL serializer routed by tscircuit — tscircuit

Antmicro GMSL serializer routed by Freerouting — Freerouting

Antmicro GMSL serializer by Antmicro · Apache-2.0 · source

Seven99.1%200timed out100%0012,062100%02554013092.5%1807608100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	2	0	0	timed out
TraceMaker	0	0	1	2,062
tscircuit	0	25	540	130
Freerouting	18	0	7	608
Human designs	0	0	0	–

Seven as its designers routed it — Human designs

Seven routed by Freerouting — Freerouting

Seven by ID8 Engineering AB · CERN-OHL-P-2.0 · source

Argos demo board100%00079100%00086100%018114104100%00128100%000–

router	Shorts	Rule errors	Seconds
Zap!	0	0	79
TraceMaker	0	0	86
tscircuit	18	114	104
Freerouting	0	1	28
Human designs	0	0	–

Numbers only: its licence (CERN-OHL-S) asks that modified designs be shared in full. Source.

Antmicro CVBS-to-MIPI bridge100%002181100%00079100%03224014498.1%4016283100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	2	181
TraceMaker	0	0	0	79
tscircuit	0	32	240	144
Freerouting	4	0	16	283
Human designs	0	0	0	–

Antmicro CVBS-to-MIPI bridge as its designers routed it — Human designs

Antmicro CVBS-to-MIPI bridge routed by Zap! — Zap!

Antmicro CVBS-to-MIPI bridge routed by TraceMaker — TraceMaker

Antmicro CVBS-to-MIPI bridge routed by tscircuit — tscircuit

Antmicro CVBS-to-MIPI bridge routed by Freerouting — Freerouting

Antmicro CVBS-to-MIPI bridge by Antmicro · Apache-2.0 · source

Trellis Core100%000622100%000827100%0213080100%00242100%000–

router	Shorts	Rule errors	Seconds
Zap!	0	0	622
TraceMaker	0	0	827
tscircuit	2	130	80
Freerouting	0	2	42
Human designs	0	0	–

Trellis Core as its designers routed it — Human designs

Trellis Core routed by TraceMaker — TraceMaker

Trellis Core routed by tscircuit — tscircuit

Trellis Core routed by Freerouting — Freerouting

Trellis Core by Protolux Electronics SARL-S · CERN-OHL-P-2.0 · source

Antmicro PMOD I3C sensor board100%00713100%0079197.2%5382324299.4%1186197100%007–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	7	13
TraceMaker	0	0	7	91
tscircuit	5	38	232	42
Freerouting	1	1	86	197
Human designs	0	0	7	–

Antmicro PMOD I3C sensor board as its designers routed it — Human designs

Antmicro PMOD I3C sensor board routed by Zap! — Zap!

Antmicro PMOD I3C sensor board routed by TraceMaker — TraceMaker

Antmicro PMOD I3C sensor board routed by tscircuit — tscircuit

Antmicro PMOD I3C sensor board routed by Freerouting — Freerouting

Antmicro PMOD I3C sensor board by Antmicro · Apache-2.0 · source

Ender-3 load-cell bed-levelling sensor100%00252100%00239100%01328510399.3%10237100%000–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	2	52
TraceMaker	0	0	2	39
tscircuit	0	13	285	103
Freerouting	1	0	2	37
Human designs	0	0	0	–

Ender-3 load-cell bed-levelling sensor as its designers routed it — Human designs

Ender-3 load-cell bed-levelling sensor routed by Zap! — Zap!

Ender-3 load-cell bed-levelling sensor routed by TraceMaker — TraceMaker

Ender-3 load-cell bed-levelling sensor routed by tscircuit — tscircuit

Ender-3 load-cell bed-levelling sensor routed by Freerouting — Freerouting

Ender-3 load-cell bed-levelling sensor by Lakshminarayanan · MIT · source

ECP5 BGA-381 FT2232HQ FIFO board100%000141100%00015982.6%2502663100%00035100%000–

router	Unrouted	Rule errors	Seconds
Zap!	0	0	141
TraceMaker	0	0	159
tscircuit	25	26	63
Freerouting	0	0	35
Human designs	0	0	–

ECP5 BGA-381 FT2232HQ FIFO board as its designers routed it — Human designs

ECP5 BGA-381 FT2232HQ FIFO board routed by Zap! — Zap!

ECP5 BGA-381 FT2232HQ FIFO board routed by TraceMaker — TraceMaker

ECP5 BGA-381 FT2232HQ FIFO board routed by tscircuit — tscircuit

ECP5 BGA-381 FT2232HQ FIFO board routed by Freerouting — Freerouting

ECP5 BGA-381 FT2232HQ FIFO board by gildobjanschi · MIT · source

RP2040 development board100%0042,69592.8%10053,072100%016279351887.7%17035227100%003–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	4	2,695
TraceMaker	10	0	5	3,072
tscircuit	0	162	793	518
Freerouting	17	0	35	227
Human designs	0	0	3	–

Numbers only: by decision. Source.

Antmicro signal-integrity test board100%00015495.3%50035566.3%36172034100%000198100%003–

router	Unrouted	Shorts	Rule errors	Seconds
Zap!	0	0	0	154
TraceMaker	5	0	0	355
tscircuit	36	17	20	34
Freerouting	0	0	0	198
Human designs	0	0	3	–

Antmicro signal-integrity test board as its designers routed it — Human designs

Antmicro signal-integrity test board routed by Zap! — Zap!

Antmicro signal-integrity test board routed by TraceMaker — TraceMaker

Antmicro signal-integrity test board routed by tscircuit — tscircuit

Antmicro signal-integrity test board routed by Freerouting — Freerouting

Antmicro signal-integrity test board by Antmicro · Apache-2.0 · source

Antmicro dual IPEX CSI interposer100%00872100%008991100%003561100%006824100%000–

router	Rule errors	Seconds
Zap!	8	72
TraceMaker	89	91
tscircuit	35	61
Freerouting	68	24
Human designs	0	–

Antmicro dual IPEX CSI interposer as its designers routed it — Human designs

Antmicro dual IPEX CSI interposer routed by Zap! — Zap!

Antmicro dual IPEX CSI interposer routed by TraceMaker — TraceMaker

Antmicro dual IPEX CSI interposer routed by tscircuit — tscircuit

Antmicro dual IPEX CSI interposer routed by Freerouting — Freerouting

Antmicro dual IPEX CSI interposer by Antmicro · Apache-2.0 · source

Arduino water-level monitor100%0009100%00014100%0038100%0077100%000–

router	Rule errors	Seconds
Zap!	0	9
TraceMaker	0	14
tscircuit	3	8
Freerouting	7	7
Human designs	0	–

Arduino water-level monitor as its designers routed it — Human designs

Arduino water-level monitor routed by Zap! — Zap!

Arduino water-level monitor routed by TraceMaker — TraceMaker

Arduino water-level monitor routed by tscircuit — tscircuit

Arduino water-level monitor routed by Freerouting — Freerouting

Arduino water-level monitor by Srushti D · MIT · source

Exact numbers

25 boards	Fully connectedboards ↑	Unroutedconnections ↓	Shortscount ↓	Rule errorscount ↓	Secondsmedian, all boards ↓	Longestseconds ↓	Timed outboards ↓	Connected, no errorsboards ↑	Manufacturing findingsJLCPCB limits ↓
Zap!	14	196	0	187	184	937	0	6	1,181
Zap!, 60 minutesoption minutes: 60	15	124	0	166	427	timed out	10	6	1,214
TraceMaker	7	808	0	439	timed out (14 of 25)	timed out	14	3	2,202
tscircuit60-minute limit	10	4,864	1,081	8,617	130	timed out	1	0	5,100
Freerouting	11	2,073	4	2,318	529	timed out	5	2	2,376
Human designs	25	0	0	82	–	–	–	13	3,235
Variants
Zap! previous buildserved until 2026-10-13	10	230	0	200	42	770	0	5	6,009
TraceMakerwithout pairs first	7	1,143	0	370	601	605	0	2	2,067
tscircuit10-minute limit	10	5,885	648	6,693	142	694	0	0	3,452

Manufacturing findings: Zap!'s check against JLCPCB standard limits.

Boards from the other routers' own projects

KiCad default rules (no rules file in their repositories).

board	routed by	Unrouted↓	Shorts↓	Rule errors↓	Seconds↓
KiCad demo: PIC programmer	Zap! (earlier run)	0	0	0	21
	TraceMaker (earlier 10-minute run)	4	0	0	108
	tscircuit (10 min)	0	20	72	19
	Published (human-routed)	0	0	0	–
KiCad demo: video	Zap! (earlier run)	231	0	0	699
	TraceMaker (earlier 10-minute run)	248	0	0	601
	tscircuit (10 min)	1,458	0	0	603
	Published (human-routed)	0	0	0	–
TraceMaker test board (plane + SMD)	Zap! (earlier run)	1	0	0	1
	TraceMaker (earlier 10-minute run)	1	0	0	267
	tscircuit (10 min)	0	49	262	231
	Published (published unrouted)	47	0	0	–
tscircuit Core1106 carrier	Zap! (earlier run)	0	0	1	46
	TraceMaker (earlier 10-minute run)	5	0	1	601
	tscircuit (10 min)	0	2	464	100
	Published (routed by tscircuit's own pipeline)	0	0	1,452	–
tscircuit RP2040 LiPo charger	Zap! (earlier run)	20	0	6	7
	TraceMaker (earlier 10-minute run)	4	0	6	601
	tscircuit (10 min)	0	5	258	49
	Published (routed by tscircuit's own pipeline)	0	5	510	–

Exact numbers

25 boards	Fully connectedboards ↑	Unroutedconnections ↓	Shortscount ↓	Rule errorscount ↓	Secondsmedian, all boards ↓	Longestseconds ↓	Timed outboards ↓	Connected, no errorsboards ↑	Manufacturing findingsJLCPCB limits ↓
Zap!best of attempts within 60 minutes, 6 CPU threads	18	100	0	174	746	timed out	8	6	1,298
TraceMaker	12	432	0	461	2,062	timed out	10	5	2,704
tscircuit60-minute limit	10	4,864	1,081	8,617	130	timed out	1	0	5,100
Freerouting	11	2,100	4	2,323	283	timed out	3	2	2,382
Human designs	25	0	0	82	–	–	–	13	3,235
Variants
Zap! previous buildserved until 2026-10-13	10	230	0	200	42	770	0	5	6,009
TraceMakerwithout pairs first	7	1,143	0	370	601	605	0	2	2,067
tscircuit10-minute limit	10	5,885	648	6,693	142	694	0	0	3,452

Manufacturing findings: Zap!'s check against JLCPCB standard limits.

Boards from the other routers' own projects

KiCad default rules (no rules file in their repositories).

board	routed by	Unrouted↓	Shorts↓	Rule errors↓	Seconds↓
KiCad demo: PIC programmer	Zap! (earlier run)	0	0	0	21
	TraceMaker (earlier 10-minute run)	4	0	0	108
	tscircuit (10 min)	0	20	72	19
	Published (human-routed)	0	0	0	–
KiCad demo: video	Zap! (earlier run)	231	0	0	699
	TraceMaker (earlier 10-minute run)	248	0	0	601
	tscircuit (10 min)	1,458	0	0	603
	Published (human-routed)	0	0	0	–
TraceMaker test board (plane + SMD)	Zap! (earlier run)	1	0	0	1
	TraceMaker (earlier 10-minute run)	1	0	0	267
	tscircuit (10 min)	0	49	262	231
	Published (published unrouted)	47	0	0	–
tscircuit Core1106 carrier	Zap! (earlier run)	0	0	1	46
	TraceMaker (earlier 10-minute run)	5	0	1	601
	tscircuit (10 min)	0	2	464	100
	Published (routed by tscircuit's own pipeline)	0	0	1,452	–
tscircuit RP2040 LiPo charger	Zap! (earlier run)	20	0	6	7
	TraceMaker (earlier 10-minute run)	4	0	6	601
	tscircuit (10 min)	0	5	258	49
	Published (routed by tscircuit's own pipeline)	0	5	510	–

Method

Measured 12–14 Oct 2026.
Boards: 25 public KiCad projects, open licence, human-routed. Placement kept; tracks, vias and pours removed; the board's own net classes and rules.
TraceMaker and Freerouting: 1 CPU core, up to 60 min per board, no GPU; TraceMaker used up to 25 GB, more than Zap!'s 14 GB; Freerouting's Java memory capped at 12 GB.
tscircuit: 1 CPU core, up to 60 min per board; it checks its limit only between solver steps, so a run still inside one step was stopped at 75 min. Zap! stops after a fixed amount of work on 1 CPU core: 3.1 min median, 16 min longest.
Zap!: router build re92d124, the build zap.mutual.solutions runs now; the same machines and session as the others, 1 CPU core, single run (the service default). One board, the Antmicro ECP5 DC-SCM, can't yet be routed on zap.mutual.solutions: the service runs several routing attempts at once, and on this board together they need more than its 14 GB job limit.
Zap!, 60 minutes: the upload option minutes: 60, in the same session on 1 CPU core per board; 10 of 25 boards used the full 60 minutes. The hosted service runs up to 6 of its attempts at once.
Check: KiCad 10.0.6 design-rule check after refilling pours, each board's rules; at most 199 findings of one kind per board (499 for spacing).
Fully connected: every pad that should be joined is joined.
Time: for each router, the median seconds on the boards that both it and Zap! fully connected within 3,600 s, beside Zap!'s median on those same boards; the board view lists every board's time.
Timed out: a run over 3,600 s; its time reads timed out, its other numbers are scored from the board it left.
Unrouted: pad pairs that should be joined but are not (KiCad's count).
Short: copper of two nets touching.
Errors: design-rule errors, shorts included; 74 per router come from the placement.
Time: start to finished board file.
Colour: Zap! cyan, others grey; board view log10(1 + value) against the column's largest value.
Race: FTDI JTAG programmer, a dense board where every router finished; shown as one of the 25.
Race, Zap!: copper per net in the order of a traced re-run, byte-identical to the measured run, its times scaled to the measured run; final shapes.
Race, others: TraceMaker, tscircuit, Freerouting: in-between states recorded from runs of the same versions and settings, final board identical to the benchmark, times scaled to the benchmark run.
Race clock: 6 times real time until Zap! finishes, then moved by scrolling. Unrouted starts at KiCad's count on the unrouted board (260). Number colours: best of the four cyan, worst coral, log scale.

Versions and settings

Zap!	Zap! router build re92d124.57335f59, the build zap.mutual.solutions runs now; boards routed 2026-10-14	the same machines and session as the others, 1 CPU core per board, single run (the service default); no clock: stops after a fixed amount of work (3.1 min median, 16 min longest here); makes its own copper pours
Zap!, 60 minutesoption minutes: 60	Zap! router build re92d124.57335f59, the build zap.mutual.solutions runs now; boards routed 2026-10-14	the same machines and session as the others, 1 CPU core per board: the single run, then more attempts one after another (twice and four times the work, then other router settings) until one has every connection and no short, or 60 minutes pass; keeps the best
TraceMaker	TraceMaker v0.9.0-116-g49fbd3c (main branch, commit 49fbd3c of 2026-10-14, the latest public version when run), built CPU-only with GCC 14.2; boards routed 2026-10-14	--time 3600 --threads 1 --no-kb --diff-pairs --keep-vias-off-pads --vias-off-pads-below 1000; 1 CPU core per board, no GPU
tscircuit60-minute limit	@tscircuit/capacity-autorouter 0.0.973 (AutoroutingPipelineSolver9_PreloadedTraceGraph) on Node.js v24.21.0, through this page's runner from the corrected input conversion; boards routed 2026-10-14 (UTC)	effort 1 (tscircuit's default); 1 CPU core per board; solver limit 3,600 s (60 minutes, the Zap! service's job limit), stopped at 75 minutes at the latest; a board stopped at the limit has no routing, since tscircuit returns routing only after its last stage
Freerouting	Freerouting 2.5.0 (latest stable release, published 2026-10-05; tag v2.5.0, build revision b647a2f824cfbc54bc4dcaee71b03b7cd945eb76 from the jar manifest), routed through the Specctra DSN and SES files that KiCad's Freerouting plugin uses (exported and imported with KiCad 10.0.6); boards routed 2026-10-14	1 CPU core, 60-minute job limit, 12 GB Java memory cap, its own defaults otherwise
Zap! previous buildserved until 2026-10-13	Zap! router build r009e659.8f766f9f, served until service release 2026.10.13.29 on 2026-10-13; boards routed 2026-10-12	same settings as the served build
TraceMakerwithout pairs first	TraceMaker v0.9.0-90-gc6fcb7b (an earlier build, commit of 2026-10-11); boards routed 2026-10-13	--time 600 --threads 1, without --diff-pairs (TraceMaker's own default pair handling)
tscircuit10-minute limit	@tscircuit/capacity-autorouter 0.0.973 (AutoroutingPipelineSolver9_PreloadedTraceGraph) on Node.js v24.21.0; boards routed 2026-10-13 (UTC) from the corrected input conversion	effort 1 (tscircuit's default); solver stopped at 600 s. tscircuit routes in stages and returns routing only after its last stage, so a board stopped at the limit has no routing at all: it never returns a partial result

Where tscircuit wrote no routing

cause	boards
Error: its pre-check reported a connection as impossible, but the designers' own route for it fits the converted board	2
Limit: a fixed step limit inside one solver stage (it ran out of iterations)	1
Error after 1442 s: repair04 merge rejected a new via in the boundary collar	1
Killed at the 75-minute backstop, past its 60-minute time limit, while still inside one solver step; no routing written	1

On tscircuit's own benchmark

problem set	Problems	Zap!verified %	tscircuitverified %	Zap!solver says %	tscircuitsolver says %
dataset01	85	100.0	100.0	100.0	100.0
srj18	15	100.0	86.7	100.0	100.0
srj19	200	71.0	92.0	71.0	93.5
srj20	200	51.5	79.0	51.5	79.0
srj21	10	100.0	100.0	100.0	100.0
srj23	76	100.0	97.4	86.8	98.7

Run 2026-10-13. tscircuit's checker for both; verified = every requested connection complete. Zap!: build 009e659 (served until 2026-10-13), JLC minimum via (0.4 mm; the problems use 0.3 mm). tscircuit: @tscircuit/capacity-autorouter 0.0.976 at 1256328 (commit of 2026-10-12), MIT licence. srj18 without one problem that overlaps our held-out boards.

Reproduce

# TraceMaker 49fbd3c, CPU-only build, on 1 CPU core
tracemaker route in.kicad_pcb -o routed.kicad_pcb --time 3600 --threads 1 --no-kb --diff-pairs \
    --keep-vias-off-pads --vias-off-pads-below 1000

# Freerouting 2.5.0 on 1 CPU core, on the DSN that KiCad exports
java -Xmx12g -Xms256m -jar \
  freerouting-2.5.0.jar -de in.dsn -do \
  out.ses --gui.enabled=false --api_server.enabled=false --profile.allow_telemetry=false \
  --usage_and_diagnostic_data.disable_analytics=true --router.max_threads=1 --router.job_timeout=01:00:00 --router.result_json=result.json

# tscircuit: @tscircuit/capacity-autorouter 0.0.973 on 1 CPU core through this page's runner,
# effort 1, 60-minute limit, stopped at 75 minutes at the latest

# Zap! through the service (API key: sign in at /auth/login and create one on your account page)
python3 zap.py run in.kicad_pcb

# check any result
kicad-cli pcb drc --refill-zones --severity-all --format json -o drc.json routed.kicad_pcb

Data: compare.json, written 2026-10-14 15:17:52 UTC. Questions or a board to add: hello@mutual.solutions.

Method

Measured 12–14 Oct 2026.
Boards: 25 public KiCad projects, open licence, human-routed. Placement kept; tracks, vias and pours removed; the board's own net classes and rules.
TraceMaker and Freerouting: 6 CPU threads, up to 60 min per board, no GPU; Freerouting's Java memory capped at 12 GB.
Runs on 14 October (UTC): TraceMaker and Freerouting 09:30 to 10:40, Zap! 12:35 and 16:14 to 17:15; all on the same machine type, with 6 consecutive vCPUs per board.
Peak memory per board: TraceMaker 60 GB, Freerouting 8 GB, Zap! 69 GB, the sum of its attempts running at once.
tscircuit uses one thread: 1 CPU core, up to 60 min per board; it checks its limit only between solver steps, so a run still inside one step was stopped at 75 min. Zap! keeps the best of its attempts within 60 minutes on 6 CPU threads: 12.4 min median, 8 of 25 boards timed out.
Zap!: router build re92d124 on 6 CPU threads per board: up to 6 routing attempts at once with different settings of the build, new ones as threads free up, until one has every connection and no short (the running ones then finish) or 60 minutes pass; keeps the best. One board, the Antmicro ECP5 DC-SCM, can't yet be routed on zap.mutual.solutions: the service runs several routing attempts at once, and on this board together they need more than its 14 GB job limit.
Check: KiCad 10.0.6 design-rule check after refilling pours, each board's rules; at most 199 findings of one kind per board (499 for spacing).
Fully connected: every pad that should be joined is joined.
Time: for each router, the median seconds on the boards that both it and Zap! fully connected within 3,600 s, beside Zap!'s median on those same boards; the board view lists every board's time.
Timed out: a run over 3,600 s; its time reads timed out, its other numbers are scored from the board it left.
Unrouted: pad pairs that should be joined but are not (KiCad's count).
Short: copper of two nets touching.
Errors: design-rule errors, shorts included; 74 per router come from the placement.
Time: start to finished board file.
Colour: Zap! cyan, others grey; board view log10(1 + value) against the column's largest value.
Race: FTDI JTAG programmer, a dense board where every router finished; shown as one of the 25.
Race, Zap!: copper per net in the order of a traced re-run, byte-identical to the measured run, its times scaled to the measured run; final shapes.
Race, others: TraceMaker, tscircuit, Freerouting: in-between states recorded from runs of the same versions and settings, final board identical to the benchmark, times scaled to the benchmark run.
Race clock: 6 times real time until Zap! finishes, then moved by scrolling. Unrouted starts at KiCad's count on the unrouted board (260). Number colours: best of the four cyan, worst coral, log scale.

Versions and settings

Zap!best of attempts within 60 minutes, 6 CPU threads	Zap! router build re92d124.57335f59, the build zap.mutual.solutions runs now; boards routed 2026-10-14	6 CPU threads per board: up to 6 routing attempts at once with different settings of the build, new ones as threads free up, until one has every connection and no short (the running ones then finish) or 60 minutes pass; keeps the best (12.4 min median, 8 of 25 boards timed out here); makes its own copper pours
TraceMaker	TraceMaker v0.9.0-116-g49fbd3c (main branch, commit 49fbd3c of 2026-10-14, the latest public version when run), built CPU-only with GCC 14.2; boards routed 2026-10-14	--time 3600 --threads 6 --no-kb --diff-pairs --keep-vias-off-pads --vias-off-pads-below 1000; 6 CPU threads per board, no GPU
tscircuit60-minute limit	@tscircuit/capacity-autorouter 0.0.973 (AutoroutingPipelineSolver9_PreloadedTraceGraph) on Node.js v24.21.0, through this page's runner from the corrected input conversion; boards routed 2026-10-14 (UTC)	effort 1 (tscircuit's default); 1 CPU core per board; solver limit 3,600 s (60 minutes, the Zap! service's job limit), stopped at 75 minutes at the latest; a board stopped at the limit has no routing, since tscircuit returns routing only after its last stage
Freerouting	Freerouting 2.5.0 (latest stable release, published 2026-10-05; tag v2.5.0, build revision b647a2f824cfbc54bc4dcaee71b03b7cd945eb76 from the jar manifest), routed through the Specctra DSN and SES files that KiCad's Freerouting plugin uses (exported and imported with KiCad 10.0.6); boards routed 2026-10-14	6 CPU threads, 60-minute job limit, 12 GB Java memory cap, its own defaults otherwise
Zap! previous buildserved until 2026-10-13	Zap! router build r009e659.8f766f9f, served until service release 2026.10.13.29 on 2026-10-13; boards routed 2026-10-12	same settings as the served build
TraceMakerwithout pairs first	TraceMaker v0.9.0-90-gc6fcb7b (an earlier build, commit of 2026-10-11); boards routed 2026-10-13	--time 600 --threads 1, without --diff-pairs (TraceMaker's own default pair handling)
tscircuit10-minute limit	@tscircuit/capacity-autorouter 0.0.973 (AutoroutingPipelineSolver9_PreloadedTraceGraph) on Node.js v24.21.0; boards routed 2026-10-13 (UTC) from the corrected input conversion	effort 1 (tscircuit's default); solver stopped at 600 s. tscircuit routes in stages and returns routing only after its last stage, so a board stopped at the limit has no routing at all: it never returns a partial result

Where tscircuit wrote no routing

cause	boards
Error: its pre-check reported a connection as impossible, but the designers' own route for it fits the converted board	2
Limit: a fixed step limit inside one solver stage (it ran out of iterations)	1
Error after 1442 s: repair04 merge rejected a new via in the boundary collar	1
Killed at the 75-minute backstop, past its 60-minute time limit, while still inside one solver step; no routing written	1

On tscircuit's own benchmark

problem set	Problems	Zap!verified %	tscircuitverified %	Zap!solver says %	tscircuitsolver says %
dataset01	85	100.0	100.0	100.0	100.0
srj18	15	100.0	86.7	100.0	100.0
srj19	200	71.0	92.0	71.0	93.5
srj20	200	51.5	79.0	51.5	79.0
srj21	10	100.0	100.0	100.0	100.0
srj23	76	100.0	97.4	86.8	98.7

Reproduce

# TraceMaker 49fbd3c, CPU-only build, on 6 CPU threads
tracemaker route in.kicad_pcb -o routed.kicad_pcb --time 3600 --threads 6 --no-kb --diff-pairs \
    --keep-vias-off-pads --vias-off-pads-below 1000

# Freerouting 2.5.0 on 6 CPU threads, on the DSN that KiCad exports
java -Xmx12g -Xms256m -jar \
  freerouting-2.5.0.jar -de in.dsn -do \
  out.ses --gui.enabled=false --api_server.enabled=false --profile.allow_telemetry=false \
  --usage_and_diagnostic_data.disable_analytics=true --router.max_threads=6 --router.job_timeout=01:00:00 --router.result_json=result.json

# tscircuit: @tscircuit/capacity-autorouter 0.0.973 on 1 CPU core through this page's runner,
# effort 1, 60-minute limit, stopped at 75 minutes at the latest

# Zap! through the service (API key: sign in at /auth/login and create one on your account page)
python3 zap.py run in.kicad_pcb --minutes 60

# check any result
kicad-cli pcb drc --refill-zones --severity-all --format json -o drc.json routed.kicad_pcb

Data: compare.json, written 2026-10-14 17:23:57 UTC. Questions or a board to add: hello@mutual.solutions.

Boards and licences

board	by	licence	source
Antmicro CM4 baseboard LVDS adapter	Antmicro	Apache-2.0	4af945e
Antmicro CVBS-to-MIPI bridge	Antmicro	Apache-2.0	7cbcec2
Antmicro dual IPEX CSI interposer	Antmicro	Apache-2.0	6b6bd0a
Antmicro ECP5 DC-SCM	Antmicro	Apache-2.0	0019313
Antmicro GMSL serializer	Antmicro	Apache-2.0	58641df
Antmicro PMOD I3C sensor board	Antmicro	Apache-2.0	5ee5d34
Antmicro signal-integrity test board	Antmicro	Apache-2.0	de0b592
Antmicro STM32H7 HDMI SO-DIMM board	Antmicro	Apache-2.0	27e3ac9
Arduino water-level monitor	Srushti D	MIT	7b47f46
Argos demo board	CNES	CERN-OHL-S-2.0	1eccb3f
Bitaxe Ultra (fork)	BastelPichi, fork of skot/bitaxe	CERN-OHL-S-2.0	0c3e71f
Buildbotics controller	Buildbotics LLC	CERN-OHL-S-2.0	1a2f36d
discretesim test board	cnlohr	MIT	f7a169e
ECP5 BGA-381 FT2232HQ FIFO board	gildobjanschi	MIT	81403f7
EmberOne	256 Foundation	CERN-OHL-S-2.0	b21b32c
Ender-3 load-cell bed-levelling sensor	Lakshminarayanan	MIT	eae5c6f
FTDI JTAG programmer	Anthony Bernaert	CERN-OHL-P-2.0	c33005e
HackRF Pro	Great Scott Gadgets	CERN-OHL-P-2.0	359b98f
Kintex PCIe board	Piotr Binkowski	Apache-2.0	056792c
RP2040 development board	acon	MIT	d943286
RP2350-PICO2-BB48	Olimex Ltd	MIT	74b0303
Seven	ID8 Engineering AB	CERN-OHL-P-2.0	7c69a14
Tiliqua (apfaudio)	Sebastian Holzapfel	CERN-OHL-S-2.0	4ee5878
Trellis Core	Protolux Electronics SARL-S	CERN-OHL-P-2.0	db4fe71
Xerxes	Rapid Analysis	Apache-2.0	43abf57

No routing images for CERN-OHL-S boards (the licence asks that modified designs be shared in full). Licences checked 2026-10-13.

Connect your agent

Introducing Introducing

Your rules, in full

tscircuit: 18 rule types

Freerouting: 9 rule types

TraceMaker: 1 rule type

Every measure

Boards from the other routers' own projects

Boards from the other routers' own projects

Versions and settings

Where tscircuit wrote no routing

On tscircuit's own benchmark

Reproduce

Versions and settings

Where tscircuit wrote no routing

On tscircuit's own benchmark

Reproduce

Introducing