BDX Bench

AI coding models tested on one prompt. Real playable builds. Transparent scoring.

Round

Oct 2026

Models

9

Audited builds

11

Shared prompts

1

Audit dimensions

5

Showdown Score v2Re-audited at implementation level, not by keyword scan: 5 dimensions × 20 points, zero for absent features. Every verified defect stays published. How scores are produced

Featured builds

11 builds, one prompt. Every card runs the real artifact in place — play it, then read the score it earned.

Ranked by Showdown Score v2

ember-dead2026-09-26

6 enemy types with genuine role separation, not stat sticks

EMBER DEAD

Space Bunny Free

91.00of 100
cont
comb
cont
audi
poli
3 findingsTouch ready
pyre-burn-horde2026-09-18

Five interlocking fire systems: heat, ignite, contagion, ground pools, death chains

PYRE — Burn the Horde

DeepSeek V4.1 Flash

80.00of 100
cont
comb
cont
audi
poli
4 findings · 2 failingNo touch supportalt build 61.0
claude-haiku-5-52026-10-08

Flamethrower cone: 175 px range, angled hit test, 42 DPS direct burn

INFERNO — Zombie Burner

Claude Haiku 5.5

63.00of 100
cont
comb
cont
audi
poli
7 findingsNo touch support
firebreak-night-shift2026-09-18

4 enemy types with real behavioural divergence read inside the AI

Firebreak: Night Shift

GPT Luna 5.6

62.00of 100
cont
comb
cont
audi
poli
4 findings · 2 failingTouch ready
inferno-dead2026-09-26

Genuinely working touch layer, verified via CDP multi-touch synthesis

INFERNO DEAD

DeepSeek V4.1 Flash

61.00of 100
cont
comb
cont
audi
poli
4 findings · 2 failingTouch readyalt build 61.0
cinderline2026-09-22

Proven continuous flame cone, measured 16.5 to 27.8 degrees with range

Cinderline

GPT 6 Sol

58.00of 100
cont
comb
cont
audi
poli
4 findings · 2 failingTouch ready
pyro-vs-zombies2026-09-17

Real contagion cascades plus tank death explosions and 12% chain reactions

PYRO vs ZOMBIES

Muse Spark 1.3

52.00of 100
cont
comb
cont
audi
poli
4 findings · 1 failingTouch ready
emberfall2026-09-22

Real continuous cone with a correctly wrapped angle helper

Emberfall

GPT Luna 6

51.00of 100
cont
comb
cont
audi
poli
4 findings · 1 failingTouch ready
space-bunny2026-09-23

A genuine radar minimap with real-time enemy and obstacle tracking

Space Bunny (superseded)

Space Bunny Free

49.00of 100
cont
comb
cont
audi
poli
4 findings · 3 failingNo touch support
pyroclasm-inferno2026-09-17

A real 1/2/3 weapon system with three distinct mechanics

PYROCLASM: Zombie Inferno

Gemini 3.8 Flash

43.00of 100
cont
comb
cont
audi
poli
5 findings · 5 failingNo touch support
zombie-fire-survival2026-09-25

A well-tuned burn DoT with four secondary channels: slow, char, smoke, recolour

Zombie Fire Survival

Gemini Pro Agent

24.00of 100
cont
comb
cont
audi
poli
5 findings · 4 failingNo touch support

Audit credibility

Every Showdown Score v2 number comes from reading the build's source and playing it, not from scanning strings. Defects are published, not filtered out.

Implementation-level audit

Builds audited

11

Verified defects

48

Failing

22

Partial pass

18

Known issue

8

How we score

5 dimensions × 20 points = 100. A feature earns points only when it is implemented and reachable. On-screen strings, comments, and dead code score zero.

  • Controls & Mobility14.4 / 20
  • Combat Physics & Weapons12.0 / 20
  • Content & Enemy Variety8.8 / 20
  • Audio & Sound Design10.7 / 20
  • Visual Polish & Game Feel11.7 / 20

Bars = cohort mean · ≥16 pass · 10–15 partial pass · <10 failure

Read the full methodology

Highest-severity findings

One verified failure per build, weakest-scoring build first. A benchmark that hides these is not worth reading.

Absolute zero audio

Failure

6/6 probes empty and 0 AudioContext constructions at runtime. The game is completely silent.

Zombie Fire SurvivalGemini Pro Agent24/100

Advertised dash does not exist

Failure

The menu says Move & Dash; there is no dash, no i-frames and no shift handling.

PYROCLASM: Zombie InfernoGemini 3.8 Flash43/100

No flamethrower implementation

Failure

The word flamethrower appears exactly once in the whole file: the page title. There is no cone, no arc, no range code.

Space Bunny (superseded)Space Bunny Free49/100

One enemy archetype

Failure

A single Zombie class with a size slider and no type field or AI branch.

EmberfallGPT Luna 651/100

Per-build audit ledger: score, verified defect count, failing count, and the worst verified defect for every audited build.
BuildModelScoreFindingsFailingWorst verified defect
EMBER DEADSpace Bunny Free9130No roguelite upgrade layerPartial pass
PYRE — Burn the HordeDeepSeek V4.1 Flash8042Zero touch supportFailure
INFERNO — Zombie BurnerClaude Haiku 5.56370No touch inputPartial pass
Firebreak: Night ShiftGPT Luna 5.66242Reachable pause hangFailure
INFERNO DEADDeepSeek V4.1 Flash6142All four enemy types share one AIFailure
CinderlineGPT 6 Sol5842Knockback entirely absentFailure
PYRO vs ZOMBIESMuse Spark 1.35241touchcancel leaves the flamethrower latchedFailure
EmberfallGPT Luna 65141One enemy archetypeFailure
Space Bunny (superseded)Space Bunny Free4943No flamethrower implementationFailure
PYROCLASM: Zombie InfernoGemini 3.8 Flash4355Advertised dash does not existFailure
Zombie Fire SurvivalGemini Pro Agent2454Absolute zero audioFailure

Severity labels follow the audit contract: failing = the build cannot do the thing it advertises · partial pass = implemented but wrong · known issue = cosmetic or unreachable. Each build's full finding list, including every line of evidence, sits on its card in Featured builds.

Global benchmark telemetry

Dataset scope
Models tracked

9

Verified weights

Benchmarks

1

Standard mini suites

Evaluation runs

9

Automated & manual

Providers

3

AI gateways & labs

Latest round

Sep 2026

Methodology v1.0

Counts reflect the current dataset: 9 models · 1 benchmark · 3 providers.

Top models

Ranked by Showdown Score v2. Dimension columns are the best-scoring audited build for each model, scored out of 20.

Open full leaderboard
Models ranked by Showdown Score, with all five audit dimensions and the verified defect count for each model's best build.
#ModelProviderShowdownControlsCombatContentAudioPolishFindings
1Space Bunny Free2 builds auditedopencode91.0of 100
19
18
17
19
18
3no failures
2DeepSeek V4.1 Flash2 builds auditedbdx-ai80.0of 100
16
17
14
17
16
42 failing
3Claude Haiku 5.5claude-relay63.0of 100
15
16
7
12
13
7no failures
4GPT Luna 5.6bdx-ai62.0of 100
15
13
11
12
11
42 failing
5GPT 6 Solbdx-ai58.0of 100
15
11
9
11
12
42 failing
6Muse Spark 1.3bdx-ai52.0of 100
12
12
9
8
11
41 failing
7GPT Luna 6bdx-ai51.0of 100
15
9
6
10
11
41 failing
8Gemini 3.8 Flashbdx-ai43.0of 100
10
9
6
9
9
55 failing
9Gemini Pro Agentbdx-ai24.0of 100
11
6
2
0
5
54 failing

Bar colour is the audit status for that dimension: lime = pass, amber = partial pass, red = failure. Findings are verified defects, published in full on each build. Other dimensions show as Not evaluated until measured.

Intelligence vs. price

Pricing and speed were not measured for these builds. See status

Not measured in this round

The eight verified builds were evaluated for game-build quality only. Price and speed charts will appear once vendor pricing and throughput measurements are published.

Compare verified builds

Category leaders

Zombie Flamethrower Showdown · game-build · the only category with measured data.

11 builds · 9 models

Showdown Score · category leader

Space Bunny Free

EMBER DEAD · 2 builds audited · 3 verified findings

91.0of 100

Controls

Space Bunny Free

19/20

3 of 11 builds pass

Combat

Space Bunny Free

18/20

3 of 11 builds pass

Content

Space Bunny Free

17/20

1 of 11 builds pass

Audio

Space Bunny Free

19/20

2 of 11 builds pass

Polish

Space Bunny Free

18/20

2 of 11 builds pass

Leaders are the highest-scoring audited build per dimension, out of 20 points each. Other categories show as Not evaluated until measured.

Latest models

Claude Haiku 5.5

claude-relay · release date not published

Play this build

Muse Spark 1.3

bdx-ai · release date not published

Play this build

Gemini 3.8 Flash

bdx-ai · release date not published

Play this build

DeepSeek V4.1 Flash

bdx-ai · release date not published

Play this build

Capability trend

Single snapshot — trends need at least two rounds. Open trends

September 2026: 94 · 93 · 91 · 88 · 86 · 82 · 79 · 55

DeepSeek V4.1 Flash leads at 94.0, followed by Space Bunny Free at 93.0, GPT 6 Sol at 91.0, GPT Luna 5.6 at 88.0, Gemini 3.8 Flash at 86.0, Muse Spark 1.3 at 82.0, GPT Luna 6 at 79.0, and Gemini Pro Agent at 55.0. The next round will draw the first trend line.

Benchmark coverage

All benchmarks

Zombie Flamethrower Showdown

coding · Showdown Score 0-100

8 evaluated builds · Showdown Score (manual game-build evaluation)

How we measure

Small, focused rounds with the same brief, clear checks, and judge review. Read the full rules before citing a number.

Read methodology →
  • Fixed suite

    Same brief and playability checks for every build in a round.

  • Hands-on scoring

    Showdown Scores from direct review: 92 for Muse Spark 1.3, 88 for Gemini 3.8 Flash.

  • Open results

    Per-build notes, provenance, and play links on every profile.

BDX Bench

Independent benchmarks, pricing, speed, capability and model intelligence in one place.

BDX Bench is an independent evaluation project. Scores on this page cover the September 2026 Zombie Flamethrower Showdown round. See Methodology for scoring details.