Model comparison
GPT-5.4 nano vs Inkling
Inkling is the stronger model overall, scoring 44.1 to 41.9 on the Noometry Index. GPT-5.4 nano costs 5.6× less per token, which makes it the better buy when Inkling's lead doesn't matter for your workload.
Last verified . 33 shared benchmarks.
Summary
- They share 33 benchmarks with published results for both. GPT-5.4 nano scores higher in 2 categories and Inkling in 6 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Inkling leads 40.4 to 23.7.
- The biggest single-benchmark swing is ARC-AGI-2: 5.7% for GPT-5.4 nano and 36.5% for Inkling.
- GPT-5.4 nano is cheaper at $0.20 / $1.25 per million input/output tokens, against $1.87 / $4.68 for Inkling.
- GPT-5.4 nano accepts more context: 400K tokens versus 66K.
- Inkling has downloadable open weights; the other is API-only.
Side by side
| GPT-5.4 nano | Inkling | |
|---|---|---|
| Provider | OpenAI | Thinking Machines Lab |
| Noometry Index | 41.9 | 44.1 |
| Released | 2026-03-17 | 2026-07-15 |
| Weights | Proprietary | Open |
| Context window | 400K | 66K |
| Max output | 128K | 66K |
| Input $ / M tokens | $0.20 | $1.87 |
| Output $ / M tokens | $1.25 | $4.68 |
| Results tracked | 40 | 41 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 nano leads
GPT-5.4 nano: 43.6 (#84), Inkling: 34.5 (#234)
| Benchmark | GPT-5.4 nano | Inkling |
|---|---|---|
| SciCode | 46.9% | 47% |
| WeirdML | 49.2% | 32.3% |
| LMArena Coding | 1405 | 1464 |
| ALE-Bench | 1,005 | 946 |
| FrontierCode | — | 14% |
| LMArena WebDev | — | 1413 |
| FrontierSWE | — | 4.1% |
Agentic & Tool Use Not comparable
GPT-5.4 nano: —, Inkling: 29.6 (#85)
| Benchmark | GPT-5.4 nano | Inkling |
|---|---|---|
| APEX-Agents | — | 33.8% |
| τ²-bench Banking | — | 25% |
Reasoning Inkling leads
GPT-5.4 nano: 23.7 (#173), Inkling: 40.4 (#56)
| Benchmark | GPT-5.4 nano | Inkling |
|---|---|---|
| ARC-AGI-2 | 5.7% | 36.5% |
| ARC-AGI-1 | 51.5% | 79.5% |
| CritPt | 9.3% | 5.4% |
| Chess Puzzles | 30% | 21% |
| LMArena Hard Prompts | 1381 | 1451 |
| DTBench | 80.3% | 87.5% |
| LMCA | 36.9% | 37.6% |
| Epoch Capabilities Index | 145.81 | 148.54 |
| SimpleBench | — | 50% |
| Kagi LLM Benchmark | 39.7% | — |
| Mystery Game Puzzles | 9% | — |
| ForecastBench | 57.3 | — |
Math GPT-5.4 nano leads
GPT-5.4 nano: 40.9 (#88), Inkling: 31.3 (#225)
| Benchmark | GPT-5.4 nano | Inkling |
|---|---|---|
| FrontierMath (Tiers 1-3) | 44.9% | 33.3% |
| FrontierMath Tier 4 | 12.2% | 4.9% |
| OTIS Mock AIME 2024-2025 | 87.8% | 88.9% |
| ProofBench | 5% | 0% |
| LMArena Math | 1406 | 1479 |
| FrontierMath (Feb 2025 set) | 25.9% | — |
| FrontierMath Tier 4 (v1) | 6.3% | — |
Knowledge Inkling leads
GPT-5.4 nano: 41.9 (#103), Inkling: 55.1 (#49)
| Benchmark | GPT-5.4 nano | Inkling |
|---|---|---|
| GPQA Diamond | 78.5% | 88.3% |
| SimpleQA Verified | 11.7% | 40.3% |
| LMArena Expert | 1396 | 1465 |
| Vectara Hallucination Rate | 3.1% | — |
Multimodal Not comparable
GPT-5.4 nano: 36.7 (#78), Inkling: —
| Benchmark | GPT-5.4 nano | Inkling |
|---|---|---|
| LMArena Vision | 1196 | — |
Multilingual Inkling leads
GPT-5.4 nano: 48.6 (#140), Inkling: 54.0 (#52)
| Benchmark | GPT-5.4 nano | Inkling |
|---|---|---|
| LMArena Non-English | 1359 | 1434 |
| LMArena Chinese | 1392 | 1490 |
| LMArena French | 1396 | 1458 |
| LMArena German | 1367 | 1446 |
| LMArena Japanese | 1343 | 1429 |
| LMArena Korean | 1320 | 1404 |
| LMArena Russian | 1363 | 1429 |
| LMArena Spanish | 1371 | 1448 |
Instruction Following Inkling leads
GPT-5.4 nano: 71.9 (#144), Inkling: 75.1 (#71)
| Benchmark | GPT-5.4 nano | Inkling |
|---|---|---|
| LMArena Instruction Following | 1362 | 1426 |
Long Context Inkling leads
GPT-5.4 nano: 41.6 (#137), Inkling: 43.8 (#86)
| Benchmark | GPT-5.4 nano | Inkling |
|---|---|---|
| LMArena Longer Query | 1366 | 1434 |
Writing & Preference Inkling leads
GPT-5.4 nano: 55.7 (#142), Inkling: 65.2 (#51)
| Benchmark | GPT-5.4 nano | Inkling |
|---|---|---|
| LMArena Text | 1372 | 1441 |
| LMArena Creative Writing | 1314 | 1387 |
| LMArena Multi-Turn | 1382 | 1436 |
| EQ-Bench Creative Writing | — | 1611 |
| EQ-Bench 4 | — | 1226 |
Frequently asked questions
Is GPT-5.4 nano better than Inkling?
Inkling is the stronger model overall, scoring 44.1 to 41.9 on the Noometry Index. GPT-5.4 nano costs 5.6× less per token, which makes it the better buy when Inkling's lead doesn't matter for your workload.
Which is cheaper, GPT-5.4 nano or Inkling?
GPT-5.4 nano is cheaper. It lists at $0.20 per million input tokens and $1.25 per million output tokens; Inkling lists at $1.87 and $4.68.
Is GPT-5.4 nano or Inkling better for coding?
GPT-5.4 nano scores higher on coding benchmarks: 43.6 versus 34.5 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 nano does, with 400K tokens against 66K.
How many benchmarks do GPT-5.4 nano and Inkling share?
33 benchmarks have published results for both models. GPT-5.4 nano has 40 scored results on Noometry and Inkling has 41.