
Opus 4.8 vs Inkling
The reasoning king — deepest thinking, premium price. vs A 975B open-weights frontier model — yours to own and run.
Head-to-head verdict: Opus 4.8 wins 41–6.
What I tested — same prompt, two models
I run the same fixed prompt set through every new model the day it drops — same string, one shot, single HTML file out — and I score the result 0–10 on whether it ran, how close it hit the brief, and how good it looked. Below is what came out when I gave the exact same prompts to Opus 4.8 and Inkling, side by side, on 47 shared tasks inside the Agent Operating System.
Both models were given identical prompts inside the Agent Operating System — no help, no iteration, no "best of N" tricks. I run each prompt once, save the HTML file the model produces, and score it 0–10 on whether it ran, how close it hit the brief, and how good it looked. The scoring is mine. The verdicts below are pulled from my source comparison guides at agentos.guide where I publish every score and the reasoning behind it.
Opus 4.8 · The default when the build has to ship on the first prompt — Opus is the safety net inside Agent OS for hard one-shots.
Inkling · Benched on GoldieBench one-shot through Tinker's OpenAI-compatible endpoint at medium reasoning effort, then headless-playtested on the same rubric as the whole field. In the Agent OS it's wired into the opencode tab on your own Tinker key — the Ink Machine.
Side-by-side on 50 shared tasks
Click any cell to play that model's actual one-shot attempt. Medals are derived from my 0–10 scores per task (highest = 🥇, second = 🥈, third = 🥉).
Where Opus 4.8 beat Inkling
The tasks where I gave Opus 4.8 a higher 0–10 score on the same prompt — with the actual commentary from my source guides.
What I saw: 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 instea…
What I saw: 5KB · plays clean · webgl, rAF
What I saw: 27KB · plays clean · three, webgl
What I saw: 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…
What I saw: 5KB · plays clean · webgl, rAF
Where Inkling beat Opus 4.8
The tasks where I gave Inkling a higher 0–10 score on the same prompt — with the actual commentary from my source guides.
What I saw: Renders cleanly with polished dock, desktop icons, and a functional Terminal window with prompt; drag, close/minimize dots, Paint canvas and localStorage Notes all present per source. Weak point: the title banner is partially hidden behind the window and the empty terminal body l…
What I saw: Renders a vivid, colorful WebGL aurora curtain with clean green/cyan/purple bands, ground plane, and tasteful title overlay — clearly on-brief and polished; but the curtain reads as a contained rectangular slab rather than a sky-spanning flowing veil, and the mountain silhouette …
What I saw: Gorgeous dense glyph rain with katakana/symbol mix, glowing trails, and an Orbitron title that reads beautifully; the hue-shifting green-to-blue gradient is striking but drifts slightly from canonical Matrix green, keeping it just shy of the top.
What I saw: Renders a clean 3D table with rack, pockets, and a decent title overlay, but the table sits small in the frame with heavy vignette wasting most of the screen, and the physics have flaws (cue-ball-only drag, no game rules/scoring). The rack looks slightly offset and the balls are …
What I saw: Renders cleanly with polished neon 3D visuals, glowing gradient title, grid arena, and functional falling-orb catch mechanic. Strong presentation but gameplay is shallow/generic — no lives, misses, difficulty ramp, or lose condition, so it lands short of the task's best entries.
Strengths & weaknesses I logged
Opus 4.8
Strengths
- Most consistent across the Goldie Bench bench — no weak build, 8.46/10 average
- Deepest one-shot reasoning, especially on game-feel and physics
- Extended thinking mode handles up to 1M tokens of context
Trade-offs
- 5–10× the per-token cost of every other model on the bench
- Less flair on cinematic visuals than GLM-5.2 — playing it safer wins on accuracy, costs you on showpiece moments
Inkling
Strengths
- Genuinely open-weights — the full 975B model is public on Hugging Face; run it on your own key, no black box
- Best one-shot builds are 2D / animation / web — a matrix-rain that topped its task (8.4), plus arcade, fractal, aurora and a mini web-OS all judged shippable (7.6–8.2)
- Frontier-class agentic coding for an open model — 77.6% SWE-bench Verified, ahead of Nemotron 3 Ultra
- 1M-token context, native multimodal (text/image/audio), and a controllable thinking-effort dial
Trade-offs
- One-shot 3D games are weak — three.js dungeons/racers render a title screen but no playable scene, like most open models (crypt 2.5)
- Physics and particle sims are hit-or-miss — black-hole, plasma and cloth one-shots often render dark or static (2.3–3.5)
- Not the strongest overall — the closed frontier (Fable 5) still tops the raw benchmarks; Inkling trades peak for ownership
Pricing & context — the spec sheet
| Spec | Opus 4.8 | Inkling |
|---|---|---|
| Vendor | Anthropic | Thinking Machines |
| Context window | 200,000 tokens (1M with extended thinking) | 1,000,000 tokens |
| Price | $15 / $75 per M tokens | $0.33 / M |
| Pricing detail | Premium pricing via the Anthropic API: $15 per million input tokens, $75 per million output tokens. Extended thinking is included but adds latency. | Inkling is open-weights — a 975B-parameter (41B active) Mixture-of-Experts model whose full weights are public on Hugging Face. You run it on your own key through Tinker's OpenAI-compatible endpoint (usage-based, ~$0.33/M sampling, 50% off at launch), or via Together / Fireworks / Modal / Databricks / Baseten. Benched here one-shot at medium reasoning effort via Tinker. |
| Release | 2026-05 | 2026-07 |
| Bench coverage | 47/47 scored · avg 7.51/10 | 50/50 scored · avg 6.07/10 |
The verdict — which should you pick?
Across 47 scored shared tasks, Opus 4.8 averaged 7.51/10, beating Inkling's 6.00/10 by 1.51 points. Pick Opus 4.8 when the build has to ship on the first prompt and you can afford the trade-offs in the comparison below.
If you only run one of these inside your stack, the head-to-head average above is the call. If you can run both, my honest play is to wire Opus 4.8 and Inkling both into the Agent Operating System and dispatch each from the kanban by task type — mission-critical one-shot builds where 'has to work the first time' matters → Opus 4.8, owning a frontier model instead of renting one — on your own key, pennies per build → Inkling. That's the same setup I run for the 4,000+ founders inside the AI Profit Boardroom.
FAQ — Opus 4.8 vs Inkling
Which is better, Opus 4.8 or Inkling?
On Goldie Bench, Opus 4.8 averages 7.51/10 across the shared tasks, with 3 gold, 1 silver, 6 bronze overall. Inkling averages 6.00/10, with 0 gold, 3 silver, 1 bronze. Opus 4.8 wins the head-to-head 41–6.
How much does Opus 4.8 cost vs Inkling?
Opus 4.8: Premium pricing via the Anthropic API: $15 per million input tokens, $75 per million output tokens. Extended thinking is included but adds latency. Inkling: Inkling is open-weights — a 975B-parameter (41B active) Mixture-of-Experts model whose full weights are public on Hugging Face. You run it on your own key through Tinker's OpenAI-compatible endpoint (usage-based, ~$0.33/M sampling, 50% off at launch), or via Together / Fireworks / Modal / Databricks / Baseten. Benched here one-shot at medium reasoning effort via Tinker.
What's the context window for Opus 4.8 vs Inkling?
Opus 4.8 has a 200,000 tokens (1M with extended thinking) context window. Inkling has a 1,000,000 tokens context window.
When should I pick Opus 4.8 over Inkling?
Pick Opus 4.8 for: Mission-critical one-shot builds where 'has to work the first time' matters; Hard reasoning tasks (planning, multi-step) where you'll pay for the depth; Anything where vendor reliability beats the per-token bill. The trade-off is the weaknesses we logged on the bench: 5–10× the per-token cost of every other model on the bench; Less flair on cinematic visuals than GLM-5.2 — playing it safer wins on accuracy, costs you on showpiece moments.
When should I pick Inkling over Opus 4.8?
Pick Inkling for: Owning a frontier model instead of renting one — on your own key, pennies per build; Generative visuals, data-viz and single-file web builds you want one-shot; A customizable open base you can fine-tune on Tinker for your own domain. The trade-off is the weaknesses we logged on the bench: One-shot 3D games are weak — three.js dungeons/racers render a title screen but no playable scene, like most open models (crypt 2.5); Physics and particle sims are hit-or-miss — black-hole, plasma and cloth one-shots often render dark or static (2.3–3.5); Not the strongest overall — the closed frontier (Fable 5) still tops the raw benchmarks; Inkling trades peak for ownership.
How does Goldie Bench score Opus 4.8 vs Inkling?
Every demo on this page was built by Julian Goldie inside the Agent Operating System — same fixed prompt for both models, one shot, single HTML file out. Each result gets a 0–10 score on whether it ran, how close it hit the brief, and how good it looked. The highest score on each task gets gold; second gets silver; third gets bronze. See methodology for full provenance.
Related comparisons
Other head-to-heads using the same scoring system:
Opus 4.8 vs Fusion Inkling vs Fusion Opus 4.8 vs Hermes MoA Inkling vs Hermes MoA Opus 4.8 vs GPT-5.6 Sol Inkling vs GPT-5.6 Sol Opus 4.8 vs Claude Fable 5 Inkling vs Claude Fable 5Full model pages: Opus 4.8 · Inkling · back to the leaderboard
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.














































