GoldieBench deep dive
Kimi K2.7's full benchmark breakdown.
The heavy lifter — frontier coder at flat-rate. Every number below is either one of my own judged one-shot builds - playable on this site - or an outside result with its source linked. Nothing display-only, nothing vibes.
01 · The headline numbers
02 · Every benchmark, bar by bar
All 25 scored tasks, best first. Hover for the judge's comment; click through for every model's build on that task. Or overlay other models in the interactive graphs.
03 · Where it wins - with the judge's own words
“All three are genuinely good. Kimi's is the jaw-dropper — a deep rainbow plunge into a seahorse spiral, dense with self-similar detail. Opus zooms smoothly into the seahorse valley with a tasteful cycling palette. GLM frames the whole iconic set in a fire palette with a live coordinate HUD, then descends. Kimi takes this one on raw spectacle.”
“All three are real, playable shooters. Opus drops you in a corridor with an imp dead ahead — gun, crosshair and HUD framed like a screenshot. Kimi matches it: a monster down a textured hall, health, ammo, minimap. GLM ships a gorgeous 'HAZARD PROTOCOL' title screen with a working game behind it, though it too spawns facing a wall. Opus by a hair on the cleanest fight.”
“Kimi nailed it — brick walls, a checkered floor, a clean minimap, textbook Wolfenstein, runs clean out of the box. Opus's is close and more atmospheric: warm fog and a vignette down a stone corridor (A/D to turn, W/S to move). GLM's engine is genuinely good — brick and mossy-stone walls, fog, a minimap — but its one-shot spawned the player buried inside a wall, dead on arrival; I nudged the start one cell so you can actually walk it. That spawn bug is why it scores lowest here, even though the e”
04 · Where it struggles - quoted, not hidden
Boards that hide the weak rows are brochures. These are Kimi K2.7's lowest scored builds, verdicts unedited:
“GLM filled the bowl with glowing liquid that actually sloshes — the most convincing 'liquid in a bowl'. Opus's particles glowed but clumped to the centre. Kimi's collapsed into a tiny blob.”
“GLM built the densest, most detailed city — windowed skyscrapers, a speed + coins HUD. Opus ran the furthest with the cleanest motion (Score 303). Kimi's runner plays fine but is unforgiving — it crashes within seconds.”
“Opus nailed the brief — labelled planet orbits, a real NEO / close-pass panel, a sim clock. GLM went for drama: a glowing nebula swirl that's gorgeous but reads more galaxy than orbit map. Kimi's is accurate but dim and sparse.”
“Opus nailed it — a pure-black event horizon, a bright photon ring, and the disk bent up and over the top exactly like the film's lensing. GLM came in strong with a clean ring and a starfield warping past the hole. Kimi's disk is fine, but the background is a soft grey blur instead of stars. This one's Opus's.”
05 · Category breakdown vs the whole field
Games (12 tasks)
Others (0 tasks)
Pages (3 tasks)
Sims (7 tasks)
Visuals (3 tasks)
06 · Outside signals - every row sourced
My bench measures one thing: judged one-shot builds. These outside rows measure other things - kept separate, never blended into the GoldieBench average, every value linked to where it comes from.
Vendor-published numbers
| Benchmark | Result | Source |
|---|---|---|
| Kimi Code Bench v2 | 62.0 (+21.8% over K2.6) | forum.moonshot.ai |
| MCP-Atlas | 76.0 (K2.6: 69.4) | forum.moonshot.ai |
| MCP Mark Verified | 81.1 (K2.6: 72.8) | forum.moonshot.ai |
| Token efficiency | reasoning token usage cut ~30% vs K2.6 | devops.com |
| Verification status | all published K2.7 numbers are Moonshot proprietary benchmarks - no independent public-suite results at release | kingy.ai |
From the source guides
| Measure | Result | Source guide |
|---|---|---|
| SWE-bench Verified (K2.6 — K2.7 internal) | 80.2% | /three-dragons |
| Cost | Pennies per build | /three-dragons |
07 · Head-to-head records
08 · Methodology + honest limits
Every GoldieBench score is a one-shot, single-file build from an identical prompt - no retries, no hand-fixing - rendered for real, screenshotted, and scored 0-10 by one judge model on one rubric across the whole field. Failures score as failures. What this bench does NOT measure: multi-turn agent work, long-context recall, or API latency - that is what the sourced outside rows are for. Full method: /methodology.
Frequently asked questions
01How does Kimi K2.7 perform on GoldieBench?
Kimi K2.7 averages 7.46/10 across 25 scored one-shot build tasks, ranking #13 of 17 ranked frontier models, with 1 task golds, 2 silvers and 0 bronzes. Every score is a real judged build you can open and play on this site.
02What is Kimi K2.7 best at?
Its strongest scored build is Fractal at 9.0/10. By category it averages Game 8.0, Page 7.2, Sim 7.1, Visual 6.7 on the bench.
03How much does Kimi K2.7 cost?
Available on Moonshot's flat-rate subscription plan — no per-token billing for individual builders. The plan covers all three speed modes (Fast, No-Think, Quality). Vendor: Moonshot AI (moonshot.ai), based in Beijing.
Source ledger
- 01Official vendor site: kimi.comkimi.com
- 02Kimi Code Bench v2forum.moonshot.ai
- 03Token efficiencydevops.com
- 04Verification statuskingy.ai
- 05SWE-bench Verified (K2.6 — K2.7 internal)agentos.guide
- 06Julian's guide: /kimi-codeagentos.guide
- 07Julian's guide: /kimi-hermesagentos.guide
- 08Julian's guide: /kimi-modes-head-to-headagentos.guide
Every model's breakdown
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.