Model comparison
GPT-5.4 vs Sakana Fugu
Head-to-head evidence from 7 shared benchmark results across 4 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.
Public leaderboard positions: GPT-5.4 #8 (Supported); Sakana Fugu unranked (Not scored). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.
Evidence parity. GPT-5.4 and Sakana Fugu share 7 comparable benchmark results. 4 of 8 categories are comparable. 45 results are unique to GPT-5.4; 4 to Sakana Fugu.
Updated July 23, 2026- Shared results
- 7
- GPT-5.4 only
- 45
- Sakana Fugu only
- 4
- Comparable categories
- 4 / 8
Treat this as a split decision. GPT-5.4 makes more sense if you need the larger 1.05M context window; Sakana Fugu is the better fit if knowledge is the priority.
Confidence note. This is a partial-evidence comparison with 7 shared benchmark results across 4 evidence categories; 4 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.
Why this result
GPT-5.4 and Sakana Fugu finish on the same BenchAlign overall score, so this is less about a single winner and more about where the edge shows up. The BenchAlign headline says tie; the benchmark table is where the real choice happens.
GPT-5.4 gives you the larger context window at 1.05M, compared with 1M for Sakana Fugu.
Category breakdown
Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.
| Category | GPT-5.4 | Δ | Sakana Fugu |
|---|---|---|---|
| Knowledge | GPT-5.457.6 | Margin→ 37.9 | Sakana Fugu95.5 |
| Multimodal | GPT-5.473.2 | Margin→ 11.9 | Sakana Fugu85.1 |
| Agentic | GPT-5.477.2 | Margin→ 3.0 | Sakana Fugu80.2 |
| Coding | GPT-5.457.7 | Margin→ 2.0 | Sakana Fugu59.7 |
| Reasoning | GPT-5.4Not measured | MarginNo overlap | Sakana Fugu86.6 |
| Math | GPT-5.442.5 | MarginNo overlap | Sakana FuguNot measured |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
Terminal-Bench 2.0
AgenticA 75.1%B 80.2%Winner: Sakana FuguΔ 5.1Terminal-Bench 2.0: GPT-5.4 scored 75.1%; Sakana Fugu scored 80.2%. Sakana Fugu wins this benchmark. - Source ↗
GPQA
KnowledgeA 92.8%B 95.5%Winner: Sakana FuguΔ 2.7GPQA: GPT-5.4 scored 92.8%; Sakana Fugu scored 95.5%. Sakana Fugu wins this benchmark. - Source ↗
CharXiv
MultimodalA 82.8%B 85.1%Winner: Sakana FuguΔ 2.3CharXiv: GPT-5.4 scored 82.8%; Sakana Fugu scored 85.1%. Sakana Fugu wins this benchmark. - Source ↗
SWE-bench Pro
CodingA 57.7%B 59%Winner: Sakana FuguΔ 1.3SWE-bench Pro: GPT-5.4 scored 57.7%; Sakana Fugu scored 59%. Sakana Fugu wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | GPT-5.4 | Sakana Fugu | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | GPT-5.4$2.5 input / $15 output | Sakana FuguNot available | A complete price comparison is not available. |
| Generation speedtokens per second | GPT-5.474 tok/s | Sakana FuguNot available | A complete speed comparison is not available. |
| First-answer latencyseconds to first token | GPT-5.4151.79 s | Sakana FuguNot available | A complete latency comparison is not available. |
| Context windowmaximum listed tokens | GPT-5.41.05M | Sakana Fugu1M | GPT-5.4 lists the larger context window. |
Benchmark Deep Dive
AgenticSakana Fugu wins17 benchmarks
| Benchmark | GPT-5.4 | Sakana Fugu | Result |
|---|---|---|---|
| Terminal-Bench 2.0Source | 75.1% | 80.2% | Sakana Fugu leads |
| CyberGymSource | 79.0% | — | Not comparable |
| BrowseCompSource | 82.7% | — | Not comparable |
| OSWorld-VerifiedSource | 75% | — | Not comparable |
| MCP AtlasSource | 70.6% | — | Not comparable |
| ToolathlonSource | 54.6% | — | Not comparable |
| τ²-bench resultsSource | 87.1% | — | Not comparable |
| Claw-EvalSource | 60.3% | — | Not comparable |
| DeepSearchQASource | 73.6% | — | Not comparable |
| AA Agentic IndexSource | 41.1% | — | Not comparable |
| APEX-Agents-AASource | 33.3% | — | Not comparable |
| GDPval-AASource | 44.7% | — | Not comparable |
| GDPval-AASource | 1395 | — | Not comparable |
| Gert LabsSource | 64.89% | — | Not comparable |
| ResearchClawBenchSource | 15.3% | — | Not comparable |
| JobBenchSource | 38.9% | — | Not comparable |
| ExploitGymSource | 6.0% | — | Not comparable |
CodingSakana Fugu wins9 benchmarks
| Benchmark | GPT-5.4 | Sakana Fugu | Result |
|---|---|---|---|
| LiveCodeBench ProSource | 87.5% | 87.8% | Sakana Fugu leads |
| SWE-bench ProSource | 57.7% | 59% | Sakana Fugu leads |
| React Native EvalsSource | 85.3% | — | Not comparable |
| Vibe Code BenchSource | 67.42% | — | Not comparable |
| AA Coding IndexSource | 71.0% | — | Not comparable |
| AA-SciCodeSource | 56.6% | — | Not comparable |
| Terminal-Bench 2.0Source | — | 80.2% | Not comparable |
| LiveCodeBench v6Source | — | 92.9% | Not comparable |
| SciCodeSource | — | 60.1% | Not comparable |
Reasoning3 benchmarks
KnowledgeSakana Fugu wins13 benchmarks
| Benchmark | GPT-5.4 | Sakana Fugu | Result |
|---|---|---|---|
| GPQASource | 92.8% | 95.5% | Sakana Fugu leads |
| HLESource | 52.1% | — | Not comparable |
| HLE w/o toolsSource | 39.8% | 47.2% | Sakana Fugu leads |
| GPQA-DSource | 92.8% | 95.5% | Sakana Fugu leads |
| HealthBench HardSource | 40.1% | — | Not comparable |
| MedXpertQA (Text)Source | 59.6% | — | Not comparable |
| Artificial Analysis Intelligence IndexSource | 51.4% | — | Not comparable |
| AA-GPQA DiamondSource | 92.0% | — | Not comparable |
| AA-HLESource | 41.6% | — | Not comparable |
| AA-Omniscience IndexSource | 5.7% | — | Not comparable |
| AA-Omniscience AccuracySource | 50.0% | — | Not comparable |
| AA-Omniscience Hallucination RateSource | 88.6% | — | Not comparable |
| HealthBench ProfessionalSource | 48.1% | — | Not comparable |
Math2 benchmarks
MultimodalSakana Fugu wins11 benchmarks
| Benchmark | GPT-5.4 | Sakana Fugu | Result |
|---|---|---|---|
| MMMU-ProSource | 81.2% | — | Not comparable |
| OfficeQA ProSource | 53.2% | — | Not comparable |
| MMMU-Pro w/ PythonSource | 82.1% | — | Not comparable |
| CharXivSource | 82.8% | 85.1% | Sakana Fugu leads |
| ERQASource | 65.4% | — | Not comparable |
| SimpleVQASource | 61.1% | — | Not comparable |
| ScreenSpot ProSource | 85.4% | — | Not comparable |
| ZeroBenchSource | 41.0% | — | Not comparable |
| MedXpertQA (MM)Source | 77.1% | — | Not comparable |
| AA-MMMU-ProSource | 78.4% | — | Not comparable |
| Design Arena WebsiteSource | 1250 | — | Not comparable |
Inst. Following1 benchmarks
| Benchmark | GPT-5.4 | Sakana Fugu | Result |
|---|---|---|---|
| AA-IFBenchSource | 73.9% | — | Not comparable |
Frequently Asked Questions (5)
Which is better, GPT-5.4 or Sakana Fugu?
GPT-5.4 and Sakana Fugu are tied on the BenchAlign overall score, so the right pick depends on which category matters most for your use case.
Which is better for knowledge tasks, GPT-5.4 or Sakana Fugu?
Sakana Fugu has the edge for knowledge tasks in this comparison, averaging 95.5 versus 57.6. Inside this category, HLE w/o tools is the benchmark that creates the most daylight between them.
Which is better for coding, GPT-5.4 or Sakana Fugu?
Sakana Fugu has the edge for coding in this comparison, averaging 59.7 versus 57.7. Inside this category, SWE-bench Pro is the benchmark that creates the most daylight between them.
Which is better for agentic tasks, GPT-5.4 or Sakana Fugu?
Sakana Fugu has the edge for agentic tasks in this comparison, averaging 80.2 versus 77.2. Inside this category, Terminal-Bench 2.0 is the benchmark that creates the most daylight between them.
Which is better for multimodal and grounded tasks, GPT-5.4 or Sakana Fugu?
Sakana Fugu has the edge for multimodal and grounded tasks in this comparison, averaging 85.1 versus 73.2. Inside this category, CharXiv is the benchmark that creates the most daylight between them.
Related Comparisons
Explore More
Choose a model with this week’s evidence
Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.
One email each week. Unsubscribe anytime.