ModelBench
Sakana logo
Models/Sakana

Fugu Ultra

Quality-first multi-agent model for hard research, analysis, and competitions

Source-linked fugu Closed weights Released 2026-06-15
API data Report
Input modalitiesT
Output modalitiesT
Input price$5/1M tokens
Output price$30/1M tokens
Context window1M
Max outputNot documented
Overview

Capabilities and specifications

✓ Reasoning Yes
✓ Tool calling Yes
✓ Structured output Yes
✓ Attachments Yes
✓ Vision input Yes
× Open weights No
Creator
Sakana
Model family
fugu
Knowledge cutoff
Not documented
License
Not documented
Release date
2026-06-15
Model ID
sakana/fugu-ultra
Inference availability

Provider pricing and availability

Provider-specific identifiers, limits, and prices per million tokens.

Report a price
Providers offering Fugu Ultra
ProviderProvider model IDContextInputOutputCache readCapabilitiesSource
Vercel AI Gateway sakana/fugu-ultra 1M $5 $30 $0.5 ReasoningToolsJSON Docs
Pioneer sakana/fugu-ultra 1M $5 $30 $0.5 ReasoningToolsJSON Docs
Sakana AI fugu-ultra-20260615 1M $5 $30 $0.5 ReasoningToolsJSON Docs
Sakana AI fugu-ultra 1M $5 $30 $0.5 ReasoningToolsJSON Docs
OpenRouter sakana/fugu-ultra 1M $5 $30 $0.5 ReasoningToolsJSON Docs
Merge Gateway sakana/fugu-ultra 1M $5 $30 - ReasoningTools Docs
EmpirioLabs AI fugu-ultra 1M $7.5 $45 $1.5 ReasoningToolsJSON Docs

Capability badges appear only when the provider catalog explicitly lists support.

Serving quality

Performance and reliability

Observed serving measurements stay separate from model specifications and are never estimated from price or model size.

Throughput Not reported

Generation speed; higher is faster.

Latency / TTFT Not reported

Time until the first output token arrives.

Uptime Not reported

Successful requests over a measured period.

Performance varies by inference provider, region, and load. Provider-level measurements appear only when a source reports them.

Published evaluations

Benchmarks

Raw results remain attached to their original source and version.

Benchmark registry
CTI-REALM - Unversioned - scoreTop 2 reported models
Fugu Ultra
69.4
Fugu
67.5
CharXiv Reasoning - Unversioned - scoreTop 2 reported models
Fugu Ultra
86.6
Fugu
85.1
GPQA Diamond - Unversioned - scoreTop 2 reported models
Fugu Ultra
95.5
Fugu
95.5
Humanity’s Last Exam - Unversioned - scoreTop 2 reported models
Fugu Ultra
50.0
Fugu
47.2
LiveCodeBench - Unversioned - scoreTop 2 reported models
Fugu Ultra
93.2
Fugu
92.9
LiveCodeBench Pro - Unversioned - scoreTop 2 reported models
Fugu Ultra
90.8
Fugu
87.8
Long Context Reasoning - Unversioned - scoreTop 2 reported models
Fugu
74.7
Fugu Ultra
73.3
MRCRv2 - Unversioned - scoreTop 2 reported models
Fugu Ultra
93.6
Fugu
86.6
SWE Bench Pro - Unversioned - scoreTop 2 reported models
Fugu Ultra
73.7
Fugu
59.0
SciCode - Unversioned - scoreTop 3 reported models
Fugu
60.1
Fugu Ultra
58.7
Qwen3.7 Max
53.5
Terminal Bench 2.1 - Unversioned - scoreTop 2 reported models
Fugu Ultra
82.1
Fugu
80.2
τ3 Banking - Unversioned - scoreTop 2 reported models
Fugu
21.7
Fugu Ultra
20.6
Catalog activity

Recent record changes

Detected changes from successful source imports.

Full change log
No recent changes detected

This record has not changed in the retained import history.

Public API

Use this record

Fetch the complete source-linked model record without an API key.

API documentation
Endpoint
GET https://www.modelbench.lol/api/v1/models/sakana/fugu-ultra
curl
curl "https://www.modelbench.lol/api/v1/models/sakana/fugu-ultra"