Model comparison
GPT-4 vs Inkling-Small
Inkling-Small is the stronger model overall, scoring 46.5 to 29.1 on the Noometry Index.
Last verified . 23 shared benchmarks.
Summary
- They share 23 benchmarks with published results for both. GPT-4 scores higher in 0 categories and Inkling-Small in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Inkling-Small leads 45.1 to 10.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 1.1% for GPT-4 and 90% for Inkling-Small.
- Inkling-Small is cheaper at $0.45 / $1.20 per million input/output tokens, against $30 / $60 for GPT-4.
- Inkling-Small accepts more context: 524K tokens versus 8K.
- Inkling-Small has downloadable open weights; the other is API-only.
Side by side
| GPT-4 | Inkling-Small | |
|---|---|---|
| Provider | OpenAI | Thinking Machines Lab |
| Noometry Index | 29.1 | 46.5 |
| Released | 2023-03-14 | 2026-07-15 |
| Weights | Proprietary | Open |
| Context window | 8K | 524K |
| Max output | 8K | 1.05M |
| Input $ / M tokens | $30 | $0.45 |
| Output $ / M tokens | $60 | $1.20 |
| Results tracked | 38 | 33 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Inkling-Small leads
GPT-4: 31.6 (#283), Inkling-Small: 43.6 (#85)
| Benchmark | GPT-4 | Inkling-Small |
|---|---|---|
| LMArena Coding | 1254 | 1451 |
| LMArena WebDev | — | 1409 |
| SciCode | — | 48.7% |
| WeirdML | 12.4% | — |
| BigCodeBench Instruct | 46% | — |
| BigCodeBench Complete | 57.2% | — |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, Inkling-Small: —
| Benchmark | GPT-4 | Inkling-Small |
|---|---|---|
| METR Time Horizons | 36.1% | — |
Reasoning Inkling-Small leads
GPT-4: 17.8 (#289), Inkling-Small: 38.6 (#63)
| Benchmark | GPT-4 | Inkling-Small |
|---|---|---|
| Chess Puzzles | 4% | 18% |
| LMArena Hard Prompts | 1241 | 1423 |
| Mystery Game Puzzles | 12% | 6% |
| Epoch Capabilities Index | 125.89 | 150.15 |
| ARC-AGI-2 | — | 40.1% |
| ARC-AGI-1 | — | 84% |
| CritPt | — | 8.3% |
| DTBench | 62.7% | — |
| LMCA | 17.1% | — |
| BIG-Bench Hard | 75.1% | — |
| ForecastBench | 57.8 | — |
| HellaSwag | 95.3% | — |
| WinoGrande | 87.5% | — |
Math Inkling-Small leads
GPT-4: 10.8 (#309), Inkling-Small: 45.1 (#77)
| Benchmark | GPT-4 | Inkling-Small |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.1% | 90% |
| LMArena Math | 1269 | 1459 |
| FrontierMath (Tiers 1-3) | — | 46.3% |
| FrontierMath Tier 4 | — | 17.1% |
| ProofBench | — | 6% |
| MATH Level 5 | 23% | — |
| GSM8K | 92% | — |
Knowledge Inkling-Small leads
GPT-4: 18.4 (#282), Inkling-Small: 48.2 (#77)
| Benchmark | GPT-4 | Inkling-Small |
|---|---|---|
| GPQA Diamond | 35.7% | 88.5% |
| LMArena Expert | 1211 | 1442 |
| SimpleQA Verified | — | 19.1% |
| MMLU | 86.4% | — |
| TriviaQA | 84.8% | — |
Multimodal Not comparable
GPT-4: —, Inkling-Small: 39.1 (#62)
| Benchmark | GPT-4 | Inkling-Small |
|---|---|---|
| LMArena Vision | — | 1235 |
Multilingual Inkling-Small leads
GPT-4: 40.6 (#215), Inkling-Small: 51.7 (#104)
| Benchmark | GPT-4 | Inkling-Small |
|---|---|---|
| LMArena Non-English | 1246 | 1402 |
| LMArena Chinese | 1242 | 1465 |
| LMArena French | 1283 | 1436 |
| LMArena German | 1251 | 1405 |
| LMArena Japanese | 1209 | 1405 |
| LMArena Korean | 1184 | 1363 |
| LMArena Russian | 1251 | 1391 |
| LMArena Spanish | 1261 | 1428 |
Instruction Following Inkling-Small leads
GPT-4: 65.3 (#222), Inkling-Small: 73.8 (#114)
| Benchmark | GPT-4 | Inkling-Small |
|---|---|---|
| LMArena Instruction Following | 1241 | 1399 |
Long Context Inkling-Small leads
GPT-4: 37.7 (#212), Inkling-Small: 42.7 (#118)
| Benchmark | GPT-4 | Inkling-Small |
|---|---|---|
| LMArena Longer Query | 1244 | 1401 |
Writing & Preference Inkling-Small leads
GPT-4: 34.9 (#268), Inkling-Small: 59.6 (#107)
| Benchmark | GPT-4 | Inkling-Small |
|---|---|---|
| LMArena Text | 1263 | 1414 |
| LMArena Creative Writing | 1244 | 1331 |
| EQ-Bench Creative Writing | 752 | 1491 |
| LMArena Multi-Turn | 1257 | 1418 |
Frequently asked questions
Is GPT-4 better than Inkling-Small?
Inkling-Small is the stronger model overall, scoring 46.5 to 29.1 on the Noometry Index.
Which is cheaper, GPT-4 or Inkling-Small?
Inkling-Small is cheaper. It lists at $0.45 per million input tokens and $1.20 per million output tokens; GPT-4 lists at $30 and $60.
Is GPT-4 or Inkling-Small better for coding?
Inkling-Small scores higher on coding benchmarks: 43.6 versus 31.6 in the Noometry coding category.
Which has the bigger context window?
Inkling-Small does, with 524K tokens against 8K.
How many benchmarks do GPT-4 and Inkling-Small share?
23 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Inkling-Small has 33.