Get the Agent OS + join 4,000+ founders inside the AI Profit Boardroom → Join AIPB ($59/mo)
DJLougen (Qwen3.6-27B base · 10-trace fine-tune)

Qwable 5 27B Coder

Qwen3.6-27B with a 10-trace 'Fable 5' fine-tune — MLX-only, and good only because the base is good.

ContextQwen3.6-27B base · MLX 4-bit
PricingFree · runs locally (MLX only)
Tasks tested41
Avg score7.14/10 average
Medals🥇0 🥈0 🥉1
Release2026-06
Official vendor source
Qwable 5 27B Coder is built by DJLougen (Qwen3.6-27B base · 10-trace fine-tune) — see the vendor's own product page, pricing, and docs at huggingface.co/DJLougen/Qwable-5-27B-Coder.
Visit huggingface.co/DJLougen/Qwable-5-27B-Coder →

What is Qwable 5 27B Coder?

Qwable 5 27B Coder is the DJLougen (Qwen3.6-27B base · 10-trace fine-tune) frontier model with a Qwen3.6-27B base · MLX 4-bit context window, released 2026-06. Tagline: Qwen3.6-27B with a 10-trace 'Fable 5' fine-tune — MLX-only, and good only because the base is good.. Official source: huggingface.co/DJLougen/Qwable-5-27B-Coder.

Pricing detail. Qwable-5-27B-Coder is a full fine-tune of Qwen3.6-27B on just 10 traces (5 Fable 5 + 5 Kimi 2.7 Coder), trained in ~3 minutes. The author deliberately withheld benchmarks ('impressive numbers without methodology is part of what this release is criticizing') and calls it a cautionary tale about AI hype — explicitly NOT recommended as a production coder. It can't run in Ollama or llama.cpp (qwen35 arch unsupported); only Apple MLX runs it. The MLX 4-bit is ~15GB.

How I use it inside the Agent OS. Wired into the Agent OS Local engine via mlx_lm.server (Apple MLX) — the only runtime that loads its qwen35 arch. Scored here on the real builds it produced one-shot.

What I built with Qwable 5 27B Coder

Every model on Goldie Bench gets the same fixed prompt set — one shot, single HTML file out — and I score the result 0–10 inside the Agent Operating System. Here's what Qwable 5 27B Coder shipped on the bench: 41 one-shot demos across Qwen3.6-27B base · MLX 4-bit of context. Of those, 41 are scored against the field with my honest 0–10 from the source guides at agentos.guide.

Strengths

  • Strong concrete builds — shipped two polished, working 3D games (a dragon-flight + a synthwave racer) and a frontier-tier landing page, one prompt each
  • Runs the newest Qwen3.6 arch that Ollama/llama.cpp can't load yet — Apple MLX (native Apple Silicon) is the only way
  • Free + 100% local once downloaded; the 27B base is genuinely capable

Trade-offs

  • Slow on a Mac — ~16–21 tok/s (27B 4-bit, M4 Max), ~5–8 min per build; a heavy thinking model that over-thinks (burned 900 tokens reasoning a one-sentence answer)
  • Abstract visuals flop — its 3D particle 'wormhole' rendered as a black screen
  • The 'Qwable Coder' branding adds nothing measurable — quality is Qwen3.6-27B's; the 10-trace, 3-minute fine-tune is, by the author's own admission, a hype cautionary tale

Best for

  • Concrete, well-specified games and web pages where the strong 27B base shines
  • Running a bleeding-edge model locally that Ollama/llama.cpp can't load yet (via MLX)
  • Cost-zero local work when you don't need speed

Every demo by Qwable 5 27B Coder

41 live demos, sorted by category. Click any tile to play the actual one-shot result. Verdicts and 0–10 scores are pulled from the source guides where I posted them publicly.

Qwable 5 27B Coder one-shot build of Crypt — GoldieBench AI benchmark screenshot▶ LIVE
Crypt
Game
Strong, atmospheric title screen with glowing serif type, coffin glyphs, and a peek at the 3D pillar behind the overlay that promises a real torch-lit crawler; the menu polish is excellent but we only see the start screen, so the actual dungeon/gameplay quality is unverified and the overlay layout is a touch generic.
Qwable 5 27B Coder one-shot build of Dragonrealm — GoldieBench AI benchmark screenshot▶ LIVE
Dragonrealm
Game
A polished, atmospheric frozen-frontier scene with procedural terrain, bare snow-laden trees, distant mountains, snowfall, and clean typographic UI — renders cohesively and reads on-brief. Holds back from winner tier because there's no visible dragon in a 'dragonrealm' scene and the snowflake quads look like flat blocky sprites, so it leans pretty but generic-winter rather than uniquely dragon-themed.
Qwable 5 27B Coder one-shot build of Doom — GoldieBench AI benchmark screenshot▶ LIVE
Doom
Game
Polished, on-brand DOOM start screen with strong red glow and clean typography, and the source shows a complete raycaster with monster AI, HUD, minimap, touch controls, and win/lose states. Loses points because the screenshot only proves the menu renders—the actual gameplay isn't shown, leaving the harder visual claims unverified.
Qwable 5 27B Coder one-shot build of Racing — GoldieBench AI benchmark screenshot▶ LIVE
Racing
Game
Renders a clean 3D scene with sky shader, trees, start gate, curbs, minimap and full HUD, but the track surface is invisible/missing under the car and the curb segments float disconnected on bare grass, breaking the core racing experience. Solid framework let down by a broken track render.
Qwable 5 27B Coder one-shot build of Skyrim — GoldieBench AI benchmark screenshot▶ LIVE
Skyrim
Game
Renders a real 3D low-poly world with terrain, pine forests, a castle structure, rocks, flowers and a polished gold-on-dark Skyrim title screen with HUD bars and controls — genuinely on-brief and atmospheric. Held back from top tier by the murky/dark lighting that mutes the scene and the generic feel versus the field's best.
Qwable 5 27B Coder one-shot build of Twilightvale — GoldieBench AI benchmark screenshot▶ LIVE
Twilightvale
Game
Renders a complete 3D low-poly island with terrain, water, trees, rocks, a player character, full RPG HUD (HP/MP/XP), minimap, weather and twilight sky — a genuinely working open-world build. Weak points: the twilight atmosphere reads more flat-orange than moody, character is a crude capsule, and overall polish/lighting falls short of the field's best.
Qwable 5 27B Coder one-shot build of Voxelcraft — GoldieBench AI benchmark screenshot▶ LIVE
Voxelcraft
Game
Clean, polished start screen with a clear brief-on-message title and good control hints, and the source shows a genuine voxel engine (simplex noise terrain, chunked greedy-ish meshing, hotbar, day/night, mobile controls). But the screenshot only captures the menu — the actual voxel world never renders in the shot, so I can't verify the core gameplay, which caps it below the field's best.
Qwable 5 27B Coder one-shot build of Arcade — GoldieBench AI benchmark screenshot▶ LIVE
Arcade
Game
Clean neon 3D Breakout that renders correctly with polished HUD, start overlay, glowing paddle and 3D arena walls; the playfield perspective and lighting look genuinely good, but the brick wall isn't visible in the start state and the code is fairly standard, keeping it shippable rather than category-topping.
Qwable 5 27B Coder one-shot build of Dogfight — GoldieBench AI benchmark screenshot▶ LIVE
Dogfight
Game
Clean HUD with title, score/kills, hull bar and control hints, but the 3D scene barely renders — no visible sky gradient, terrain, or recognizable planes, just blurry orange blobs and an abstract green shape, so the dogfight world fails to come together visually.
Qwable 5 27B Coder one-shot build of Dragonflight — GoldieBench AI benchmark screenshot▶ LIVE
Dragonflight
Game
Renders a complete, atmospheric low-poly dragon flight scene with rolling terrain, snow-capped mountains, glowing crystals, towers, and a polished HUD (score/alt/speed/fire bar). The dragon model is recognizable with wings, horns and a segmented glowing tail, but the wing geometry looks somewhat crude/flat and the body proportions are a bit awkward, keeping it shippable rather than category-topping.
Qwable 5 27B Coder one-shot build of Game — GoldieBench AI benchmark screenshot▶ LIVE
Game
Game
Polished neon start screen with clean glow, HUD, and a well-structured Three.js build (detailed ship, asteroids, mobile controls), but the screenshot only shows the menu — actual gameplay polish is unverified, and as a 3D arcade shooter the concept is fairly generic versus the field's best.
Qwable 5 27B Coder one-shot build of Neonblaster — GoldieBench AI benchmark screenshot▶ LIVE
Neonblaster
Game
Clean neon HUD, starfield, and perspective grid render well with auto-fire bullets visible, but the scene is sparse — no enemies on screen, tiny indistinct ship, and the empty arena reads as underwhelming versus the polished top builds in the field.
Qwable 5 27B Coder one-shot build of Neoncity — GoldieBench AI benchmark screenshot▶ LIVE
Neoncity
Game
Strong neon-driving atmosphere with glowing cyan/magenta road strips, perspective vanishing point, overhead signs, and clean HUD/speedometer that reads as a working drive. Weak spots: buildings are very dark and undefined (windows barely visible), so the cyberpunk cityscape feels underlit and a bit sparse versus the field's best.
Qwable 5 27B Coder one-shot build of Neonracer — GoldieBench AI benchmark screenshot▶ LIVE
Neonracer
Game
Renders cleanly with a polished neon synthwave aesthetic — glowing title, neon posts, arches, lit road, and a working speed bar/score HUD all on-brief. The car body reads a bit dark and flat against the road, but overall it's a strong, atmospheric, shippable build.
Qwable 5 27B Coder one-shot build of Nordiccrypt — GoldieBench AI benchmark screenshot▶ LIVE
Nordiccrypt
Game
The HUD chrome (title, crosshair, minimap, controls) renders with a nice Nordic gold-on-dark theme and the minimap shows a real generated dungeon, but the actual 3D scene is essentially black — the crypt geometry and lighting failed to render visibly, leaving the core dungeon-crawler experience broken.
Qwable 5 27B Coder one-shot build of Outrun — GoldieBench AI benchmark screenshot▶ LIVE
Outrun
Game
Nails the Outrun/synthwave brief with a gradient retro sun, neon mountains, magenta road edges, scanline grid and a detailed 3D player car with glowing taillights — genuinely atmospheric and on-brief. Loses a touch for the mountains looking like floating spikes rather than a horizon range, but it's polished and clearly playable with working HUD/score.
Qwable 5 27B Coder one-shot build of Pool — GoldieBench AI benchmark screenshot▶ LIVE
Pool
Game
Clean 3D render with proper rack, cue ball, pockets and polished UI, but the rail geometry is visibly broken — a stray rail segment floats across the middle of the felt and the long rails are misplaced/duplicated, undercutting the table's realism.
Qwable 5 27B Coder one-shot build of Raycaster — GoldieBench AI benchmark screenshot▶ LIVE
Raycaster
Game
Renders a clean, fisheye-corrected raycast view with textured striped walls, fog shading, a glowing key sprite, and a detailed minimap showing all entities — strong and clearly on-brief. Just shy of best for slightly flat lighting and a basic stripe-only wall texture rather than richer detailing.
Qwable 5 27B Coder one-shot build of Rpg — GoldieBench AI benchmark screenshot▶ LIVE
Rpg
Game
Renders a clean, functional dungeon crawler with procedural rooms, lighting, HP/MP/XP bars, enemies, items, doors, stairs, and a polished inventory/HUD — clearly on-brief and shippable. Weak spots are the somewhat sparse/dim tile art and generic feel; solid but not quite topping the best field entries.
Qwable 5 27B Coder one-shot build of Landing — GoldieBench AI benchmark screenshot▶ LIVE
Landing
Page
Clean, polished dark AI-SaaS landing with well-crafted feature cards, gradient nav and tasteful color tokens, but the screenshot shows the hero scrolled mostly past with the Three.js particle background nearly invisible/blank, undercutting the first impression versus the field's best.
Qwable 5 27B Coder one-shot build of Webos — GoldieBench AI benchmark screenshot▶ LIVE
Webos
Page
Clean welcome overlay with polished gradient title and topbar/dock visible behind, and the source shows ambitious app implementations (notes, paint, terminal). But the screenshot is gated at the entry modal—the actual desktop OS isn't shown rendering, so it can't match the field's best on-brief presentation.
Qwable 5 27B Coder one-shot build of Blackhole — GoldieBench AI benchmark screenshot▶ LIVE
Blackhole
Sim
Strong Interstellar-style render with a convincing warped accretion disk arcing over the top and reflecting below, plus a clean starfield, well-styled UI controls and info panel; the disk wrap reads as genuine gravitational lensing rather than a flat plane, though the event horizon shadow is a bit faint/muddy and the photon ring isn't crisply defined.
Qwable 5 27B Coder one-shot build of Boids — GoldieBench AI benchmark screenshot▶ LIVE
Boids
Sim
Polished 3D boids with attractive colored cone/wing meshes, grid floor, stars, live stats, and full rule toggles plus count cycling — renders cleanly and is clearly on-brief. Falls just short of top: the flock looks fairly scattered rather than tightly emergent, and a 3D nav/orbit camera or denser clustering would push it past the best in the field.
Qwable 5 27B Coder one-shot build of Cloth — GoldieBench AI benchmark screenshot▶ LIVE
Cloth
Sim
Renders a real Verlet cloth draped over a sphere with shadows, polished UI and working controls — clearly on-brief and shippable, but the cloth mesh shows crumpled/torn-looking artifacts and the constraints don't fully resolve into a clean drape, keeping it short of the field's best.
Qwable 5 27B Coder one-shot build of Fluid — GoldieBench AI benchmark screenshot▶ LIVE
Fluid
Sim
Clean typographic overlay and the curl-noise particle system renders, but the result is a faint, sparse dust cloud rather than a convincing flowing fluid — particles read as dim points with low density and no real fluid cohesion. Polished UI but weak visual payoff against the field's best.
Qwable 5 27B Coder one-shot build of Fractal — GoldieBench AI benchmark screenshot▶ LIVE
Fractal
Sim
Renders a recognizable Mandelbrot with a clean glassy UI and full control set, but the fractal occupies only the left portion of the screen with vast empty black space — a centering/coloring bug leaves most of the canvas uncolored, hurting the visual payoff badly relative to the field's best.
Qwable 5 27B Coder one-shot build of Galaxy — GoldieBench AI benchmark screenshot▶ LIVE
Galaxy
Sim
Clean spiral-arm structure with multi-color particles, glowing core, polished UI and full interaction (drag/zoom/pan/burst/palette) — strong and shippable, but the core glow reads as a flat opaque sphere and the arms are a touch dim/sparse, keeping it just short of the field's best.
Qwable 5 27B Coder one-shot build of Orbit — GoldieBench AI benchmark screenshot▶ LIVE
Orbit
Sim
Polished UI chrome (title, sim panel, control bar) renders nicely, but the actual simulation is empty — Bodies:0, FPS:0, no rendered orbits or bodies in the canvas, so the core brief fails. The inline OrbitControls minification likely broke the render loop or body creation.
Qwable 5 27B Coder one-shot build of Particleforge — GoldieBench AI benchmark screenshot▶ LIVE
Particleforge
Sim
Renders cleanly with a polished UI, mode toggles, 20k GPU-shaded particles and a nice glowing orbital ring; weak point is the screenshot only shows a flat band/ring (the camera angle flattens the swirl) and color variety reads as washed pink/yellow rather than a striking sculptable forge, leaving it strong but not best-in-field.
Qwable 5 27B Coder one-shot build of Reactiondiff — GoldieBench AI benchmark screenshot▶ LIVE
Reactiondiff
Sim
Clean UI with full Gray-Scott controls (feed/kill/diffusion/speed), good interactivity and palette logic, but the screenshot shows mostly smooth green blobs without the crisp Turing patterns the brief implies — the palette and B-channel mapping wash out fine structure, leaving it more diffuse than diffusive.
Qwable 5 27B Coder one-shot build of Solar — GoldieBench AI benchmark screenshot▶ LIVE
Solar
Sim
Renders cleanly with a glowing sun, all 8 planets at varied sizes/colors, Saturn's rings, faint orbit ellipses, dense starfield, and a polished info panel plus speed control. Solid, shippable Three.js build—just shy of the field's best, lacking texture maps/standout flair to top the task.
Qwable 5 27B Coder one-shot build of Wormhole — GoldieBench AI benchmark screenshot▶ LIVE
Wormhole
Sim
Pretty purple particle field with a glowing central beam and clean UI, but it doesn't read as a wormhole tunnel — the camera is viewing the cylinder side-on so there's no sense of a flythrough or depth funnel, just a bokeh starfield around a vertical light shaft.
Qwable 5 27B Coder one-shot build of Aurora — GoldieBench AI benchmark screenshot▶ LIVE
Aurora
Visual
Beautiful flowing aurora ribbons with convincing additive glow, layered mountains, starfield, and clean typography—genuinely atmospheric and on-brief. Slightly undermined by the lit mountain peaks reading as flat blue triangles rather than snow caps, just shy of the top of the field.
Qwable 5 27B Coder one-shot build of Fireworks — GoldieBench AI benchmark screenshot▶ LIVE
Fireworks
Visual
Strong 3D scene with a city silhouette, starfield, rising rocket trail, and multiple live bursts with additive glow — clearly on-brief and shippable. Holds back from a winner because the explosion particles read a bit sparse/scattered rather than crisp radial bursts, and the polish is just shy of the best entries.
Qwable 5 27B Coder one-shot build of Lavalamp — GoldieBench AI benchmark screenshot▶ LIVE
Lavalamp
Visual
Convincing 3D lava lamp with a glowing glass body, metallic base, warm volumetric blobs and a tasteful title overlay — genuinely atmospheric and on-brief. Slightly let down by the blobs reading more as static glowing forms than smooth rising metaballs, and the cylinder is a bit hazy/foggy, keeping it just short of the field's best.
Qwable 5 27B Coder one-shot build of Matrix — GoldieBench AI benchmark screenshot▶ LIVE
Matrix
Visual
Authentic Matrix rain with proper katakana glyphs, bright white-green leading chars, glowing pulsing title and vignette — looks polished and dense. Strong entry but just shy of the field's best; the title glow slightly overpowers and trail rendering could be cleaner.
Qwable 5 27B Coder one-shot build of Plasma — GoldieBench AI benchmark screenshot▶ LIVE
Plasma
Visual
Classic plasma renders smoothly with rich, organic color blobs and a polished UI featuring 8 palettes, ripple interaction, and tasteful typography; the multi-octave sine field looks authentic but the buttons inherit the canvas backdrop colors awkwardly and it stops just short of the field's most striking entries.
Qwable 5 27B Coder one-shot build of Synthwave — GoldieBench AI benchmark screenshot▶ LIVE
Synthwave 🥉
Visual
Gorgeous classic synthwave scene — the banded retro sun, gradient sky, neon pillars, mountain silhouettes and reflective grid all nail the genre with polished glow and typography. Minor nit: the grid lines read more as a hazy glow than crisp perspective lines, but the overall composition is striking and clearly tops the field.
Qwable 5 27B Coder one-shot build of Terrain — GoldieBench AI benchmark screenshot▶ LIVE
Terrain
Visual
The screenshot shows only the loading screen ('Generating Terrain...') with no rendered terrain visible, indicating the 3D scene failed to display despite competent-looking source. Judged on what renders, this is effectively broken — the polished UI scaffolding and detailed simplex/biome code can't be evaluated since nothing terrain-related appeared.
Qwable 5 27B Coder one-shot build of Voxel — GoldieBench AI benchmark screenshot▶ LIVE
Voxel
Visual
Renders a clean, cohesive voxel island with proper layered terrain (stone/dirt/sand/grass), instanced trees, flowers, sky dome and a polished HUD/legend. Weakness: the framing is awkward (terrain pushed up off-center, lots of empty space, top cut off) and the camera is too high to show much of the brief's water/relief, leaving it solid but not a task winner.
Qwable 5 27B Coder one-shot build of Waves — GoldieBench AI benchmark screenshot▶ LIVE
Waves
Visual
Clean UI with title, mode buttons (Calm/Moderate/Storm) and a working shader-based ocean, but the rendered result reads as an abstract puzzle-piece/liquid blob from a near-top-down angle rather than clearly recognizable rolling ocean waves; the foam particles also look like falling snow, hurting the 'waves' read.
every demo, in a grid · click any one to play

Compare Qwable 5 27B Coder against every other model

Every head-to-head featuring Qwable 5 27B Coder. Verdicts shown for scored pairs.

Qwable 5 27B Coder vs Fusion
Fusion leads 35–6
Qwable 5 27B Coder vs Claude Opus 5
Claude Opus 5 leads 33–7
Qwable 5 27B Coder vs Hermes MoA
Hermes MoA leads 35–2
Qwable 5 27B Coder vs GPT-5.6 Sol
GPT-5.6 Sol leads 36–1
Qwable 5 27B Coder vs Claude Fable 5
Claude Fable 5 leads 31–5
Qwable 5 27B Coder vs Qwen 3.8
Qwen 3.8 leads 28–5
Qwable 5 27B Coder vs Grok
Grok leads 25–10
Qwable 5 27B Coder vs MiniMax M3
MiniMax M3 leads 27–13
Qwable 5 27B Coder vs Fugu Ultra
Fugu Ultra leads 28–12
Qwable 5 27B Coder vs Kimi K3
Kimi K3 leads 33–8
Qwable 5 27B Coder vs GLM-5.2
GLM-5.2 leads 27–14
Qwable 5 27B Coder vs Fugu Mini
Fugu Mini leads 22–12
Qwable 5 27B Coder vs Muse Spark 1.2
Muse Spark 1.2 leads 25–13
Qwable 5 27B Coder vs Opus 4.8
Opus 4.8 leads 21–20
Qwable 5 27B Coder vs Kimi K2.7
Kimi K2.7 leads 10–9
Qwable 5 27B Coder vs Gemini 3.6 Flash
Qwable 5 27B Coder leads 21–17
Qwable 5 27B Coder vs Claude Sonnet 5
Claude Sonnet 5 leads 25–15
Qwable 5 27B Coder vs Qwen 3.7
Qwable 5 27B Coder leads 28–12
Qwable 5 27B Coder vs Fugu Ultra 1.1
Fugu Ultra 1.1 leads 10–7
Qwable 5 27B Coder vs Inkling
Qwable 5 27B Coder leads 27–12
Qwable 5 27B Coder vs Grok 4.6
Qwable 5 27B Coder leads 11–5
Qwable 5 27B Coder vs Agents-A1
Qwable 5 27B Coder leads 32–8
Qwable 5 27B Coder vs Gemma 4 12B · MLX
Qwable 5 27B Coder leads 36–3
Qwable 5 27B Coder vs Laguna XS 2.1
Qwable 5 27B Coder leads 39–0
Qwable 5 27B Coder vs Qwythos 9B
Qwable 5 27B Coder leads 39–2
Qwable 5 27B Coder vs LongCat-2.0
LongCat-2.0 leads 4–0
Qwable 5 27B Coder vs Hy3
Qwable 5 27B Coder leads 2–0
Qwable 5 27B Coder vs Gemma-4 12B Coder
Qwable 5 27B Coder leads 6–0
Qwable 5 27B Coder vs DeepSeek V4 Flash
41 shared tasks · unscored
Qwable 5 27B Coder vs DeepSeek V4 Pro
41 shared tasks · unscored
Qwable 5 27B Coder vs Kimi K2.7 · Fast
41 shared tasks · unscored
Qwable 5 27B Coder vs Kimi K2.7 · No-Think
41 shared tasks · unscored
Qwable 5 27B Coder vs Kimi K2.7 · Quality
41 shared tasks · unscored
Qwable 5 27B Coder vs Ornith 1.0
41 shared tasks · unscored
Qwable 5 27B Coder vs Claude Mythos 5
Reference-only
Qwable 5 27B Coder vs Kilo Code
Reference-only

See all 66 comparisons across every model →

Quick pill index

Direct comparisons against every other scored model on the bench:

Qwable 5 27B Coder vs Fusion Qwable 5 27B Coder vs Claude Opus 5 Qwable 5 27B Coder vs Hermes MoA Qwable 5 27B Coder vs GPT-5.6 Sol Qwable 5 27B Coder vs Claude Fable 5 Qwable 5 27B Coder vs Qwen 3.8 Qwable 5 27B Coder vs Grok Qwable 5 27B Coder vs MiniMax M3 Qwable 5 27B Coder vs Fugu Ultra Qwable 5 27B Coder vs Kimi K3 Qwable 5 27B Coder vs GLM-5.2 Qwable 5 27B Coder vs Fugu Mini Qwable 5 27B Coder vs Muse Spark 1.2 Qwable 5 27B Coder vs Opus 4.8 Qwable 5 27B Coder vs Kimi K2.7 Qwable 5 27B Coder vs Gemini 3.6 Flash Qwable 5 27B Coder vs Claude Sonnet 5 Qwable 5 27B Coder vs Qwen 3.7 Qwable 5 27B Coder vs Fugu Ultra 1.1 Qwable 5 27B Coder vs Inkling Qwable 5 27B Coder vs Grok 4.6 Qwable 5 27B Coder vs Agents-A1 Qwable 5 27B Coder vs Gemma 4 12B · MLX Qwable 5 27B Coder vs Laguna XS 2.1 Qwable 5 27B Coder vs Qwythos 9B Qwable 5 27B Coder vs LongCat-2.0 Qwable 5 27B Coder vs Hy3 Qwable 5 27B Coder vs Gemma-4 12B Coder

Read more on agentos.guide:

Qwable 5 27B Coder — frequently asked

What is Qwable 5 27B Coder?

Qwable 5 27B Coder is DJLougen (Qwen3.6-27B base · 10-trace fine-tune)'s AI model — Qwen3.6-27B with a 10-trace 'Fable 5' fine-tune — MLX-only, and good only because the base is good. It has a 256K tokens context window and was released 2026-06.

How good is Qwable 5 27B Coder at coding and one-shot builds?

On the GoldieBench one-shot build benchmark it averages 7.14/10 across 41 scored tasks, with 0 gold, 0 silver and 1 bronze medals.

How much does Qwable 5 27B Coder cost?

Free · runs locally (MLX only). Qwable-5-27B-Coder is a full fine-tune of Qwen3.6-27B on just 10 traces (5 Fable 5 + 5 Kimi 2.7 Coder), trained in ~3 minutes. The author deliberately withheld benchmarks ('impressive numbers without methodology is part

Where can I see Qwable 5 27B Coder demos?

Every one-shot build is live and playable on this page and on the GoldieBench compare matrix — same prompt as every other model, no retries.

The same stack Julian uses

Run this stack yourself.

Every demo on this bench was built inside the Agent Operating System — one prompt, one shot, single HTML file out. The Agent OS, the prompts, the templates, the weekly walkthroughs and 4,000+ founders shipping with it every day all live inside the AI Profit Boardroom.

4,000+founders
258documented wins
38countries
$59/momonthly