Model comparison
GPT-5 Nano vs Inkling
Inkling is the stronger model overall, scoring 44.1 to 33.5 on the Noometry Index. GPT-5 Nano costs 19× less per token, which makes it the better buy when Inkling's lead doesn't matter for your workload.
Last verified . 31 shared benchmarks.
Summary
- They share 31 benchmarks with published results for both. GPT-5 Nano scores higher in 0 categories and Inkling in 9 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Inkling leads 65.2 to 39.1.
- The biggest single-benchmark swing is ARC-AGI-1: 20.7% for GPT-5 Nano and 79.5% for Inkling.
- GPT-5 Nano is cheaper at $0.05 / $0.40 per million input/output tokens, against $1.87 / $4.68 for Inkling.
- GPT-5 Nano accepts more context: 400K tokens versus 66K.
- Inkling has downloadable open weights; the other is API-only.
Side by side
| GPT-5 Nano | Inkling | |
|---|---|---|
| Provider | OpenAI | Thinking Machines Lab |
| Noometry Index | 33.5 | 44.1 |
| Released | 2025-08-07 | 2026-07-15 |
| Weights | Proprietary | Open |
| Context window | 400K | 66K |
| Max output | 128K | 66K |
| Input $ / M tokens | $0.05 | $1.87 |
| Output $ / M tokens | $0.40 | $4.68 |
| Results tracked | 49 | 41 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
GPT-5 Nano: 33.6 (#254), Inkling: 34.5 (#234)
| Benchmark | GPT-5 Nano | Inkling |
|---|---|---|
| WeirdML | 38.1% | 32.3% |
| LMArena Coding | 1351 | 1464 |
| ALE-Bench | 718.67 | 946 |
| FrontierCode | — | 14% |
| SWE-bench Verified (bash only) | 34.8% | — |
| LMArena WebDev | — | 1413 |
| FrontierSWE | — | 4.1% |
| SciCode | — | 47% |
Agentic & Tool Use Inkling leads
GPT-5 Nano: 25.8 (#106), Inkling: 29.6 (#85)
| Benchmark | GPT-5 Nano | Inkling |
|---|---|---|
| Terminal-Bench | 21.8% | — |
| APEX-Agents | — | 33.8% |
| Berkeley Function Calling Leaderboard | 51.5% | — |
| τ²-bench Banking | — | 25% |
Reasoning Inkling leads
GPT-5 Nano: 16.3 (#306), Inkling: 40.4 (#56)
| Benchmark | GPT-5 Nano | Inkling |
|---|---|---|
| ARC-AGI-2 | 2.6% | 36.5% |
| ARC-AGI-1 | 20.7% | 79.5% |
| Chess Puzzles | 27% | 21% |
| LMArena Hard Prompts | 1328 | 1451 |
| DTBench | 62.7% | 87.5% |
| LMCA | 7.9% | 37.6% |
| Epoch Capabilities Index | 139.38 | 148.54 |
| SimpleBench | — | 50% |
| Kagi LLM Benchmark | 62.2% | — |
| CritPt | — | 5.4% |
| Mystery Game Puzzles | 9% | — |
| ForecastBench | 59.1 | — |
Math Inkling leads
GPT-5 Nano: 29.4 (#241), Inkling: 31.3 (#225)
| Benchmark | GPT-5 Nano | Inkling |
|---|---|---|
| FrontierMath (Tiers 1-3) | 20% | 33.3% |
| FrontierMath Tier 4 | 2.4% | 4.9% |
| OTIS Mock AIME 2024-2025 | 81.1% | 88.9% |
| ProofBench | 12% | 0% |
| LMArena Math | 1317 | 1479 |
| Omni-MATH | 54.6% | — |
| MATH Level 5 | 95.2% | — |
| FrontierMath (Feb 2025 set) | 8.3% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge Inkling leads
GPT-5 Nano: 35.9 (#178), Inkling: 55.1 (#49)
| Benchmark | GPT-5 Nano | Inkling |
|---|---|---|
| GPQA Diamond | 69.4% | 88.3% |
| SimpleQA Verified | 11.7% | 40.3% |
| LMArena Expert | 1321 | 1465 |
| MMLU-Pro | 77.8% | — |
| Vectara Hallucination Rate | 10.5% | — |
| GPQA (HELM) | 67.9% | — |
Multimodal Not comparable
GPT-5 Nano: 31.3 (#108), Inkling: —
| Benchmark | GPT-5 Nano | Inkling |
|---|---|---|
| LMArena Vision | 1159 | — |
| VPCT | 37.2% | — |
Multilingual Inkling leads
GPT-5 Nano: 45.3 (#172), Inkling: 54.0 (#52)
| Benchmark | GPT-5 Nano | Inkling |
|---|---|---|
| LMArena Non-English | 1313 | 1434 |
| LMArena Chinese | 1356 | 1490 |
| LMArena German | 1327 | 1446 |
| LMArena Japanese | 1226 | 1429 |
| LMArena Korean | 1269 | 1404 |
| LMArena Russian | 1296 | 1429 |
| LMArena Spanish | 1360 | 1448 |
| LMArena French | — | 1458 |
Instruction Following Too close to call
GPT-5 Nano: 75.0 (#79), Inkling: 75.1 (#71)
| Benchmark | GPT-5 Nano | Inkling |
|---|---|---|
| LMArena Instruction Following | 1306 | 1426 |
| IFEval | 93.2% | — |
Long Context Inkling leads
GPT-5 Nano: 31.3 (#281), Inkling: 43.8 (#86)
| Benchmark | GPT-5 Nano | Inkling |
|---|---|---|
| LMArena Longer Query | 1312 | 1434 |
| Fiction.LiveBench | 44.4% | — |
Writing & Preference Inkling leads
GPT-5 Nano: 39.1 (#249), Inkling: 65.2 (#51)
| Benchmark | GPT-5 Nano | Inkling |
|---|---|---|
| LMArena Text | 1320 | 1441 |
| LMArena Creative Writing | 1249 | 1387 |
| EQ-Bench Creative Writing | 705 | 1611 |
| LMArena Multi-Turn | 1311 | 1436 |
| WildBench | 80.6% | — |
| EQ-Bench 4 | — | 1226 |
Frequently asked questions
Is GPT-5 Nano better than Inkling?
Inkling is the stronger model overall, scoring 44.1 to 33.5 on the Noometry Index. GPT-5 Nano costs 19× less per token, which makes it the better buy when Inkling's lead doesn't matter for your workload.
Which is cheaper, GPT-5 Nano or Inkling?
GPT-5 Nano is cheaper. It lists at $0.05 per million input tokens and $0.40 per million output tokens; Inkling lists at $1.87 and $4.68.
Is GPT-5 Nano or Inkling better for coding?
They score almost the same on coding (33.6 vs 34.5); test both on your own repository before choosing.
Which has the bigger context window?
GPT-5 Nano does, with 400K tokens against 66K.
How many benchmarks do GPT-5 Nano and Inkling share?
31 benchmarks have published results for both models. GPT-5 Nano has 49 scored results on Noometry and Inkling has 41.