Real head-to-head · same prompt, one shot
Hy3 vs Claude Opus 5
Tencent's open-weights coder — Apache-2.0, cheap, beats GLM-5.1 on frontend in Tencent's blind eval. vs The new Anthropic flagship — benched on all 45 one-shot builds the day it landed.
Head-to-head verdict: tied 3–3.
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 Hy3 and Claude Opus 5, side by side, on 6 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.
Hy3 · Wired into the Agent OS as the 'Hy3 Coder' tab (chat + live preview + workspace) via OpenRouter. Bench built one-shot on the same prompts as the field; weak builds iterated by Hy3 itself (the model fixes its own builds).
Claude Opus 5 · Benched on all 45 GoldieBench tasks via API on release day, incremental deploys as scores landed.
Side-by-side on 24 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 = 🥉).
Task ↓
Hy3
Claude Opus 5
Game
Game
Game
Game
Game
Game
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Page
— not attempted —
Where Hy3 beat Claude Opus 5
The tasks where I gave Hy3 a higher 0–10 score on the same prompt — with the actual commentary from my source guides.
Parachute
Game
Hy3 6.8
·
Claude Opus 5 2.0
(+4.8)
What I saw: Renders cleanly with a detailed articulated skydiver (helmet, suit, arms, boots — not a bare capsule), clean HUD with altitude/distance/phase, and a lush jungle canopy of blobs with a visible river below; but the canopy overhead reads as a flat pink slab rather than a parachute, …
Gtafoot
Game
Hy3 7.2
·
Claude Opus 5 2.5
(+4.7)
What I saw: Strong dusk-city atmosphere with a readable blocky hero (cap, jacket trim, gun), streetlights, crosswalks, and a pedestrian, plus clean HUD (ammo, health bar, wanted stars, crosshair, controls). Weak points: the world is fairly barren mid-frame, no visible buildings-as-cover in t…
Flightsim
Game
Hy3 6.8
·
Claude Opus 5 3.0
(+3.8)
What I saw: Clean HUD with working attitude indicator, heading tape, throttle/gear/score panels, and a decently modeled aircraft with wings, nav lights and prop; but the terrain reads as an empty green haze with no visible runway, trees or structures from this altitude, leaving the world fla…
Where Claude Opus 5 beat Hy3
The tasks where I gave Claude Opus 5 a higher 0–10 score on the same prompt — with the actual commentary from my source guides.
Doom
Game
Claude Opus 5 6.8
·
Hy3 4.5
(+2.3)
What I saw: Strong polished HUD, textured raycaster walls, weapon viewmodel, and a detailed minimap show real engine work, plus counters imply combat (01/10 kills, 09 demons left). But the screenshot shows NO visible enemies on screen — combat can't be confirmed in the frame, so it reads clo…
Dragonrealm
Game
Claude Opus 5 8.3
·
Hy3 7.2
(+1.1)
What I saw: Gorgeous atmospheric frozen world — aurora, moon glow, snowy terrain, low-poly pines, ruins, and a well-modeled armored third-person character with sword DRAWN and full HUD (compass, health/stam, kill counter). Strong and clearly shippable, but no visible enemies/dragon or combat…
Gtadrive
Game
Claude Opus 5 8.4
·
Hy3 7.4
(+1.0)
· polished neon driving
What I saw: Gorgeous cohesive art direction — detailed low-poly car with wheels/shadow, layered cityscape, refined HUD with integrity/velocity/wanted stars and a live minimap showing cop (red) and player markers plus traffic (purple car visible). Pursuit=2 and full control set (fire/steal/bo…
Strengths & weaknesses I logged
Hy3
Strengths
- Apache-2.0 open weights — self-host free, no lock-in
- Tencent's 270-expert blind eval: 2.67/4 vs GLM-5.1's 2.51, strongest on frontend / data / CI-CD
- Hallucination rate cut 12.5% → 5.4%; stable tool-calls across scaffoldings (<4% SWE-Bench variance)
Trade-offs
- Slow upstream on OpenRouter (30-90s per build) — fine for one-shots, sluggish for tight loops
- One-shot game builds can under-render (flat raycaster walls, unlit 3D) without an iterate pass
Claude Opus 5
Strengths
- Frontier-class coding + agentic reasoning (Claude 5 family)
- 1M-token context — reads an entire codebase in one call
- Benched here with skill-infused game prompts the day of release
Trade-offs
- Premium pricing ($5/$25 per M) — route the everyday 90% to cheaper lanes
- Reasoning-by-default eats token budgets unless tuned per call
Pricing & context — the spec sheet
| Spec | Hy3 | Claude Opus 5 |
|---|---|---|
| Vendor | Tencent Hunyuan | Anthropic |
| Context window | 262,144-token context window. Open weights (Apache-2.0) on HuggingFace / ModelScope / GitHub; benched here via OpenRouter. | 1,000,000 tokens |
| Price | $0.14 / 1M input · $0.58 / 1M output | $5 / $25 per M |
| Pricing detail | Tencent Hunyuan 3 — open-weights under Apache-2.0, so free to self-host. On OpenRouter it is one of the cheapest capable coders: ~$0.14/M in, $0.58/M out (1 RMB / 4 RMB). Upstream can be slow (30-90s to first token), but per-token cost is negligible. | Anthropic's brand-new flagship — the first Opus of the Claude 5 family, with a 1M-token context window. Benched via API the day it dropped; game tasks use our skill-infused AAA build prompts. |
| Release | 2026-07-06 | 2026-07 |
| Bench coverage | 7/7 scored · avg 6.76/10 | 23/23 scored · avg 5.07/10 |
The verdict — which should you pick?
Across 6 scored shared tasks, Hy3 averaged 6.65/10, beating Claude Opus 5's 5.17/10 by 1.48 points. Pick Hy3 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 Hy3 and Claude Opus 5 both into the Agent Operating System and dispatch each from the kanban by task type — cost-sensitive coding + frontend design where open weights matter → Hy3, hardest agentic builds → Claude Opus 5. That's the same setup I run for the 4,000+ founders inside the AI Profit Boardroom.
FAQ — Hy3 vs Claude Opus 5
Which is better, Hy3 or Claude Opus 5?
On Goldie Bench, Hy3 averages 6.65/10 across the shared tasks, with 0 gold, 0 silver, 0 bronze overall. Claude Opus 5 averages 5.17/10, with 0 gold, 0 silver, 1 bronze. It's a curated tie on the head-to-head.
How much does Hy3 cost vs Claude Opus 5?
Hy3: Tencent Hunyuan 3 — open-weights under Apache-2.0, so free to self-host. On OpenRouter it is one of the cheapest capable coders: ~$0.14/M in, $0.58/M out (1 RMB / 4 RMB). Upstream can be slow (30-90s to first token), but per-token cost is negligible. Claude Opus 5: Anthropic's brand-new flagship — the first Opus of the Claude 5 family, with a 1M-token context window. Benched via API the day it dropped; game tasks use our skill-infused AAA build prompts.
What's the context window for Hy3 vs Claude Opus 5?
Hy3 has a 262,144-token context window. Open weights (Apache-2.0) on HuggingFace / ModelScope / GitHub; benched here via OpenRouter. context window. Claude Opus 5 has a 1,000,000 tokens context window.
When should I pick Hy3 over Claude Opus 5?
Pick Hy3 for: Cost-sensitive coding + frontend design where open weights matter; Self-hosters who want an Apache-2.0 model they fully own; Anyone wiring a cheap capable coder into a live build panel (Agent OS Hy3 Coder tab). The trade-off is the weaknesses we logged on the bench: Slow upstream on OpenRouter (30-90s per build) — fine for one-shots, sluggish for tight loops; One-shot game builds can under-render (flat raycaster walls, unlit 3D) without an iterate pass.
When should I pick Claude Opus 5 over Hy3?
Pick Claude Opus 5 for: Hardest agentic builds; Whole-repo reasoning; Frontier one-shots. The trade-off is the weaknesses we logged on the bench: Premium pricing ($5/$25 per M) — route the everyday 90% to cheaper lanes; Reasoning-by-default eats token budgets unless tuned per call.
How does Goldie Bench score Hy3 vs Claude Opus 5?
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:
Hy3 vs Fusion Claude Opus 5 vs Fusion Hy3 vs Hermes MoA Claude Opus 5 vs Hermes MoA Hy3 vs GPT-5.6 Sol Claude Opus 5 vs GPT-5.6 Sol Hy3 vs Claude Fable 5 Claude Opus 5 vs Claude Fable 5Full model pages: Hy3 · Claude Opus 5 · back to the leaderboard
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