Real head-to-head · same prompt, one shot
Claude Sonnet 5 vs Grok 4.6
The agentic SWE frontier — 82% SWE-bench Verified, Dev Team mode. vs Frontier intelligence at half the frontier price.
Head-to-head verdict: Claude Sonnet 5 wins 14–4 with 2 ties.
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 Claude Sonnet 5 and Grok 4.6, side by side, on 20 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.
Claude Sonnet 5 · Reach for it in Agent OS when the job is iterative, tool-using software engineering. For one-shot visual builds, GLM 5.2 (free) beat it 4-1 here.
Grok 4.6 · Benched on 16 skill-infused game builds, every one played through its full gameplay arc before scoring, then the broken ones were handed back to Grok 4.6 to repair itself.
Side-by-side on 47 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 ↓
Claude Sonnet 5
Grok 4.6
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
— not attempted —
Game
— not attempted —
Game
— not attempted —
Page
— not attempted —
Where Claude Sonnet 5 beat Grok 4.6
The tasks where I gave Claude Sonnet 5 a higher 0–10 score on the same prompt — with the actual commentary from my source guides.
Skyrim
Game
Claude Sonnet 5 7.4
·
Grok 4.6 1.5
(+5.9)
What I saw: Clean low-poly open-world render with convincing terrain, water, trees, rocks, shadows, and dual controls for desktop/mobile — but it reads more generic 'nature explorer' than Skyrim (no snowy peaks visible, no fantasy/RPG flavor like structures, enemies, or quests), keeping it s…
Voxelcraft
Game
Claude Sonnet 5 7.2
·
Grok 4.6 1.5
(+5.7)
What I saw: Renders a working voxel world with clean hotbar, crosshair, HUD, and full FPS controls/raycast placement, but the terrain reads flat and washed-out (over-bright water plane dominating the view) and lacks the depth/polish of the best entry.
Nordiccrypt
Game
Claude Sonnet 5 6.8
·
Grok 4.6 3.0
(+3.8)
What I saw: Renders a functional first-person maze with warm torch glow, dust motes, minimap, and clean Nordic-styled UI, but the view is washed-out and flat — walls read as a beige grid without depth, stone/rune detail is barely visible, and it lacks the atmospheric contrast and dungeon men…
Neoncity
Game
Claude Sonnet 5 8.2
·
Grok 4.6 5.4
(+2.8)
What I saw: Strong on-brief cyberpunk drive: neon edge strips, cyan dashed lane markers, glowing windowed buildings and clean perspective read convincingly as a neon city. Weakness is the right side feels sparse and the buildings cluster asymmetrically, keeping it just shy of the best in field.
Racing
Game
Claude Sonnet 5 8.2
·
Grok 4.6 5.4
(+2.8)
What I saw: Strong, clean third-person render with proper checkered finish line, striped curbs, low-poly trees, HUD, and a functioning minimap showing scattered obstacles; polished and shippable but the ring track and generic box-car keep it just short of the field's best.
Where Grok 4.6 beat Claude Sonnet 5
The tasks where I gave Grok 4.6 a higher 0–10 score on the same prompt — with the actual commentary from my source guides.
Twilightvale
Game
Grok 4.6 7.4
·
Claude Sonnet 5 3.0
(+4.4)
What I saw: Moody twilight forest with a caped knight carrying a visible sword, purple crystal enemies that close on you, pines, rocks and a live minimap. Played 20s: SHARDS went 0/7 to 1/7, SCORE to 00310 and VIT fell 100 to 006 because the enemies deal damage. Real combat loop, just very p…
Gtadrive
Game
Grok 4.6 7.6
·
Claude Sonnet 5 4.5
(+3.1)
What I saw: Drivable city sandbox: multi-part orange sedan with taillights, lit tower blocks, marked roads, traffic cars, street lamps, a live minimap and a wanted-star row. Played 15s: MPH read 000 then 071 then 016 through accelerate-and-brake, so the driving model integrates properly. Lig…
Gtafoot
Game
Grok 4.6 7.4
·
Claude Sonnet 5 4.5
(+2.9)
What I saw: Third-person street level with a real armed character model, long cast shadows, a sunset skybox, pedestrians and parked cars down the block, plus a grid minimap. Played 12s: AMMO went 11 to 09 and SCORE 000010 to 000030 on click-fire, so shooting and scoring are wired. World is b…
Arcade
Game
Grok 4.6 8.7
·
Claude Sonnet 5 8.2
(+0.5)
· 3D breakout cityscape
What I saw: Strong: a fully-rendered 3D breakout with atmospheric cityscape, glowing rails, detailed paddle craft, brick grid, active ball with particle trail, and polished HUD (core/velocity/lives/score/wave). Weak: perspective makes the brick field feel small and the 3D angle slightly comp…
Strengths & weaknesses I logged
Claude Sonnet 5
Strengths
- 82.1% SWE-bench Verified — first model past 80% on real GitHub-issue repair
- Dev Team multi-agent mode + 1M context for repo-level agentic work
- Precision on hard logic — won the raycaster the open-weight field kept botching
Trade-offs
- One-shot creative-visual builds trail GLM 5.2 here (lost 4 of 5) — no iteration to catch its own bugs
- A temporal-dead-zone bug blanked its N-body orbit sim on the first shot
Grok 4.6
Strengths
- Ties GPT-5.6 Sol at 61 on the Artificial Analysis Intelligence Index, one point behind Claude Fable 5 Max
- Trained with agentic reinforcement learning for long-running agents, so it holds a spec across a 40KB single-file build
- Fixes its own broken builds: handed the exact runtime error, it repaired 4 of 4 failed games on the first retry
- Very strong arcade and shooter output - the synthwave racer and the Doom raycaster are top-tier one-shots
Trade-offs
- Flight models are its weak spot: both the flight sim and the dogfight shipped unflyable on the first pass
- Two of sixteen game builds died on a hard error (a duplicate identifier and a bad computeBoundingSphere call)
- Worlds are lit and composed but usually untextured, so terrain reads as flat coloured planes
Pricing & context — the spec sheet
| Spec | Claude Sonnet 5 | Grok 4.6 |
|---|---|---|
| Vendor | Anthropic | xAI |
| Context window | 1,000,000 tokens | 500,000 tokens |
| Price | $3 / $15 per M ($2/$10 intro) | $2 in / $6 out per M tokens |
| Pricing detail | $3.00 input / $15.00 output per million tokens; introductory $2.00/$10.00 through 2026-08-31. | xAI's August 2026 flagship. $2 per million input tokens and $6 per million output tokens on the standard tier (the Fast tier is double), which is roughly half what the other frontier models charge for the same Artificial Analysis Intelligence Index score of 61. |
| Release | 2026-06-30 | 2026-08 |
| Bench coverage | 47/47 scored · avg 7.01/10 | 20/20 scored · avg 5.95/10 |
The verdict — which should you pick?
Across 20 scored shared tasks, Claude Sonnet 5 averaged 6.96/10, beating Grok 4.6's 5.95/10 by 1.01 points. Pick Claude Sonnet 5 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 Claude Sonnet 5 and Grok 4.6 both into the Agent Operating System and dispatch each from the kanban by task type — agentic software engineering — write / run / test / fix loops on real repos → Claude Sonnet 5, high-volume agent work where the per-token bill decides what you can afford to run → Grok 4.6. That's the same setup I run for the 4,000+ founders inside the AI Profit Boardroom.
FAQ — Claude Sonnet 5 vs Grok 4.6
Which is better, Claude Sonnet 5 or Grok 4.6?
On Goldie Bench, Claude Sonnet 5 averages 6.96/10 across the shared tasks, with 0 gold, 0 silver, 4 bronze overall. Grok 4.6 averages 5.95/10, with 1 gold, 0 silver, 0 bronze. Claude Sonnet 5 wins the head-to-head 14–4.
How much does Claude Sonnet 5 cost vs Grok 4.6?
Claude Sonnet 5: $3.00 input / $15.00 output per million tokens; introductory $2.00/$10.00 through 2026-08-31. Grok 4.6: xAI's August 2026 flagship. $2 per million input tokens and $6 per million output tokens on the standard tier (the Fast tier is double), which is roughly half what the other frontier models charge for the same Artificial Analysis Intelligence Index score of 61.
What's the context window for Claude Sonnet 5 vs Grok 4.6?
Claude Sonnet 5 has a 1,000,000 tokens context window. Grok 4.6 has a 500,000 tokens context window.
When should I pick Claude Sonnet 5 over Grok 4.6?
Pick Claude Sonnet 5 for: Agentic software engineering — write / run / test / fix loops on real repos; Repo-level reasoning across a 1M-token context (Dev Team multi-agent mode); Precise logic — raycasters, physics — where one-shot open models slip. The trade-off is the weaknesses we logged on the bench: One-shot creative-visual builds trail GLM 5.2 here (lost 4 of 5) — no iteration to catch its own bugs; A temporal-dead-zone bug blanked its N-body orbit sim on the first shot.
When should I pick Grok 4.6 over Claude Sonnet 5?
Pick Grok 4.6 for: High-volume agent work where the per-token bill decides what you can afford to run; Arcade, shooter and driving builds in one shot; Self-repair loops - it is unusually good at fixing a build when you hand it the real error. The trade-off is the weaknesses we logged on the bench: Flight models are its weak spot: both the flight sim and the dogfight shipped unflyable on the first pass; Two of sixteen game builds died on a hard error (a duplicate identifier and a bad computeBoundingSphere call); Worlds are lit and composed but usually untextured, so terrain reads as flat coloured planes.
How does Goldie Bench score Claude Sonnet 5 vs Grok 4.6?
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:
Claude Sonnet 5 vs Fusion Grok 4.6 vs Fusion Claude Sonnet 5 vs Claude Opus 5 Grok 4.6 vs Claude Opus 5 Claude Sonnet 5 vs Hermes MoA Grok 4.6 vs Hermes MoA Claude Sonnet 5 vs GPT-5.6 Sol Grok 4.6 vs GPT-5.6 SolFull model pages: Claude Sonnet 5 · Grok 4.6 · 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