Real head-to-head · same prompt, one shot
Fugu Ultra vs Claude Sonnet 5
Sakana's multi-agent answer to Fusion — frontier ensemble without single-vendor risk. vs The agentic SWE frontier — 82% SWE-bench Verified, Dev Team mode.
Head-to-head verdict: Fugu Ultra wins 25–15 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 Fugu Ultra and Claude Sonnet 5, side by side, on 42 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.
Fugu Ultra · Dispatched from Agent OS as the panel-ensemble alternative to OpenRouter Fusion. Bench scored by Claude judge against the same 42 prompts as every other model.
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.
Side-by-side on 42 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 ↓
Fugu Ultra
Claude Sonnet 5
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Game
Page
Page
Sim
Sim
Sim
Where Fugu Ultra beat Claude Sonnet 5
The tasks where I gave Fugu Ultra a higher 0–10 score on the same prompt — with the actual commentary from my source guides.
Solar
Sim
Fugu Ultra 9.0
·
Claude Sonnet 5 2.5
(+6.5)
· winner · panel polish
What I saw: Fugu Ultra v2 rebuild — 55.7KB solar system, the densest solar attempt on the bench. Full </html>, animation loop, Saturn rings, drag-to-orbit + scroll-to-zoom. Smoke-test PASS (3.6% pixel diff after drag, zero console errors). The panel ensemble produces a markedly richer build …
Twilightvale
Game
Fugu Ultra 9.0
·
Claude Sonnet 5 3.0
(+6.0)
· winner · open-world depth
What I saw: Ultra v2 — 61.8KB open-world RPG (village, NPCs, weather, day/night). Smoke-test PASS. Densest Ultra build on the bench.
Aurora
Visual
Fugu Ultra 8.0
·
Claude Sonnet 5 2.5
(+5.5)
What I saw: Ultra v2 (gap-fill) — aurora over mountain ridge. Smoke-test PASS (0.8% diff — visual-only prompt).
Orbit
Sim
Fugu Ultra 8.5
·
Claude Sonnet 5 3.5
(+5.0)
What I saw: 26KB inner-solar-system orbit map with a glassmorphic info panel, kicker badge, blurred backdrop, hover cards. Cleaner UI than Fusion's same-task attempt — beats it on polish.
Wormhole
Sim
Fugu Ultra 7.0
·
Claude Sonnet 5 3.0
(+4.0)
What I saw: Ultra v2 — wormhole tunnel. Smoke-test MAYBE (0.3% diff) — animates slowly; hold-space accel didn't show big change in the short test.
Where Claude Sonnet 5 beat Fugu Ultra
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.
Fluid
Sim
Claude Sonnet 5 8.6
·
Fugu Ultra 6.5
(+2.1)
· gorgeous flow field
What I saw: Stunning rendered flow-field with rich swirling particle streaks, a clear vortex focal point, and vivid rainbow color mapping over additive-blended trails — genuinely beautiful and clearly on-brief. Only knock is the low 22fps and it's a flow-field trail sim rather than true flui…
Neonblaster
Game
Claude Sonnet 5 8.3
·
Fugu Ultra 6.5
(+1.8)
What I saw: Strong neon aesthetic with glowing nebulae, layered starfield, polished ship and auto-fire bullets, plus a full feature set (waves, bosses, power-ups, synth sequencer, screen-shake); the screenshot looks clean and on-brief but reads a touch sparse/quiet early on versus the flashi…
Pool
Game
Claude Sonnet 5 8.2
·
Fugu Ultra 6.5
(+1.7)
What I saw: Clean 3D render with proper racked triangle, numbered/striped ball textures, cue stick aiming, pockets and power bar—clearly on-brief and polished. Solid physics-oriented setup but visually generic versus a task winner, and the shadowing under the rack looks a bit off.
Outrun
Game
Claude Sonnet 5 8.6
·
Fugu Ultra 7.0
(+1.6)
· Textbook synthwave outrun
What I saw: Gorgeous, on-brief execution — striped retro sun, parallax mountains, glowing pink/cyan rumble strips and lane markers on a proper pseudo-3D road, plus a detailed neon car and polished CRT scanline/vignette overlays. Speed reads 000 in the shot (idle), but the classic Jake-Gordon…
Arcade
Game
Claude Sonnet 5 8.2
·
Fugu Ultra 7.0
(+1.2)
What I saw: Clean, polished neon Breakout that renders perfectly with glowing bricks, gradient paddle, particle bursts, and a live in-progress state (score 10, a brick already cleared). Solid physics with side-aware collisions and level scaling, but it's a well-executed take on the most comm…
Strengths & weaknesses I logged
Fugu Ultra
Strengths
- SWE Bench Pro 73.7 · GPQA-D 95.5 · MRCRv2 93.6 — Sakana's published frontier-tier benchmark scores
- Vendor-agnostic ensemble — opt out of specific providers for compliance / export-control
- OpenAI-compatible API at api.sakana.ai — drop-in for existing tooling
Trade-offs
- Panel orchestration adds latency — even a 'pong' burns ~2k orchestration tokens
- Newer than Fusion; less community calibration on long-tail prompts
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
Pricing & context — the spec sheet
| Spec | Fugu Ultra | Claude Sonnet 5 |
|---|---|---|
| Vendor | Sakana AI | Anthropic |
| Context window | 272,000 tokens with the standard rate. Calls exceeding 272K context are billed at the higher 'long-context' rates. | 1,000,000 tokens |
| Price | $5 / 1M input · $30 / 1M output (Fugu Ultra) | $3 / $15 per M ($2/$10 intro) |
| Pricing detail | Sakana's multi-agent orchestration: a single API call internally dispatches to multiple frontier models and synthesises the answer. Subscription plans run $20-$200/mo (Standard / Pro / Max); PAYG is $5/M input + $30/M output for Fugu Ultra. Direct competitor to OpenRouter Fusion's panel approach. | $3.00 input / $15.00 output per million tokens; introductory $2.00/$10.00 through 2026-08-31. |
| Release | 2026-06-15 | 2026-06-30 |
| Bench coverage | 42/42 scored · avg 7.94/10 | 42/42 scored · avg 7.18/10 |
The verdict — which should you pick?
Across 42 scored shared tasks, Fugu Ultra averaged 7.94/10, beating Claude Sonnet 5's 7.18/10 by 0.76 points. Pick Fugu Ultra 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 Fugu Ultra and Claude Sonnet 5 both into the Agent Operating System and dispatch each from the kanban by task type — teams that want fusion-class quality but need a different vendor risk profile → Fugu Ultra, agentic software engineering — write / run / test / fix loops on real repos → Claude Sonnet 5. That's the same setup I run for the 3,600+ founders inside the AI Profit Boardroom.
FAQ — Fugu Ultra vs Claude Sonnet 5
Which is better, Fugu Ultra or Claude Sonnet 5?
On Goldie Bench, Fugu Ultra averages 7.94/10 across the shared tasks, with 6 gold, 7 silver, 9 bronze overall. Claude Sonnet 5 averages 7.18/10, with 3 gold, 3 silver, 3 bronze. Fugu Ultra wins the head-to-head 25–15.
How much does Fugu Ultra cost vs Claude Sonnet 5?
Fugu Ultra: Sakana's multi-agent orchestration: a single API call internally dispatches to multiple frontier models and synthesises the answer. Subscription plans run $20-$200/mo (Standard / Pro / Max); PAYG is $5/M input + $30/M output for Fugu Ultra. Direct competitor to OpenRouter Fusion's panel approach. Claude Sonnet 5: $3.00 input / $15.00 output per million tokens; introductory $2.00/$10.00 through 2026-08-31.
What's the context window for Fugu Ultra vs Claude Sonnet 5?
Fugu Ultra has a 272,000 tokens with the standard rate. Calls exceeding 272K context are billed at the higher 'long-context' rates. context window. Claude Sonnet 5 has a 1,000,000 tokens context window.
When should I pick Fugu Ultra over Claude Sonnet 5?
Pick Fugu Ultra for: Teams that want Fusion-class quality but need a different vendor risk profile; Operators avoiding export-controlled providers (Sakana emphasises this in their pitch); Deep-research workflows where ensemble verdicts beat single-model answers. The trade-off is the weaknesses we logged on the bench: Panel orchestration adds latency — even a 'pong' burns ~2k orchestration tokens; Newer than Fusion; less community calibration on long-tail prompts.
When should I pick Claude Sonnet 5 over Fugu Ultra?
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.
How does Goldie Bench score Fugu Ultra vs Claude Sonnet 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:
Fugu Ultra vs Fusion Claude Sonnet 5 vs Fusion Fugu Ultra vs Hermes MoA Claude Sonnet 5 vs Hermes MoA Fugu Ultra vs Grok Claude Sonnet 5 vs Grok Fugu Ultra vs MiniMax M3 Claude Sonnet 5 vs MiniMax M3Full model pages: Fugu Ultra · Claude Sonnet 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 3,600+ founders shipping with it every day all live inside the AI Profit Boardroom.
3,600+founders
258documented wins
38countries
$59/momonthly