# GPT-5.6 Terra vs Inkling-Small

> GPT-5.6 Terra is the stronger model overall, scoring 59.1 to 46.5 on the Noometry Index. Inkling-Small costs 7.1× less per token, which makes it the better buy when GPT-5.6 Terra's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/gpt-5-6-terra-vs-inkling-small
- Last updated: 2026-10-10
- Shared benchmarks: 33

## Summary

- They share 33 benchmarks with published results for both. GPT-5.6 Terra scores higher in 9 categories and Inkling-Small in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.6 Terra leads 81.5 to 44.9.
- The biggest single-benchmark swing is ProofBench: 74% for GPT-5.6 Terra and 6% for Inkling-Small.
- Inkling-Small is cheaper at $0.45 / $1.20 per million input/output tokens, against $2 / $12 for GPT-5.6 Terra.
- GPT-5.6 Terra accepts more context: 1.05M tokens versus 524K.
- Inkling-Small has downloadable open weights; the other is API-only.

## Snapshot

| | GPT-5.6 Terra | Inkling-Small |
|---|---|---|
| Provider | OpenAI | Thinking Machines Lab |
| Noometry Index | 59.1 | 46.5 |
| Rank | 15 | 57 |
| Context | 1.05M | 524K |
| Input $/M | $2 | $0.45 |
| Output $/M | $12 | $1.20 |
| Weights | Proprietary | Open |

## Coding

- GPT-5.6 Terra: 57.1 (#16)
- Inkling-Small: 43.3 (#80)

| Benchmark | GPT-5.6 Terra | Inkling-Small |
|---|---|---|
| LMArena WebDev | 1520 | 1408 |
| SciCode | 55% | 48.7% |
| LMArena Coding | 1484 | 1451 |
| DeepSWE | 69.6% | — |
| FrontierCode | 41.3% | — |
| CursorBench | 41.3% | — |
| WeirdML | 78.3% | — |
| ALE-Bench | 1,951 | — |

## Agentic & Tool Use

- GPT-5.6 Terra: 40.2 (#23)
- Inkling-Small: —

| Benchmark | GPT-5.6 Terra | Inkling-Small |
|---|---|---|
| APEX-Agents | 58.2% | — |
| BALROG | 53.2% | — |
| GDP.pdf | 24.7% | — |
| Vending-Bench 2 | 7,343 | — |

## Reasoning

- GPT-5.6 Terra: 60.6 (#19)
- Inkling-Small: 39.0 (#53)

| Benchmark | GPT-5.6 Terra | Inkling-Small |
|---|---|---|
| ARC-AGI-2 | 83.9% | 40.1% |
| ARC-AGI-1 | 96.5% | 84% |
| CritPt | 30% | 8.3% |
| Chess Puzzles | 54% | 18% |
| LMArena Hard Prompts | 1467 | 1424 |
| Mystery Game Puzzles | 35% | 6% |
| Epoch Capabilities Index | 159.31 | 150.13 |
| SimpleBench | 48.9% | — |
| Kagi LLM Benchmark | 51.3% | — |
| NYT Connections (extended) | 78.4% | — |
| DTBench | 93.3% | — |
| LMCA | 55% | — |
| Surface Evolver Bench | 83.8% | — |

## Math

- GPT-5.6 Terra: 81.5 (#10)
- Inkling-Small: 44.9 (#65)

| Benchmark | GPT-5.6 Terra | Inkling-Small |
|---|---|---|
| FrontierMath (Tiers 1-3) | 86% | 46.3% |
| FrontierMath Tier 4 | 70.7% | 17.1% |
| OTIS Mock AIME 2024-2025 | 99.7% | 90% |
| ProofBench | 74% | 6% |
| LMArena Math | 1463 | 1453 |

## Knowledge

- GPT-5.6 Terra: 61.3 (#27)
- Inkling-Small: 48.3 (#71)

| Benchmark | GPT-5.6 Terra | Inkling-Small |
|---|---|---|
| GPQA Diamond | 93.3% | 88.5% |
| SimpleQA Verified | 43.2% | 19.1% |
| LMArena Expert | 1492 | 1442 |

## Multimodal

- GPT-5.6 Terra: 47.5 (#10)
- Inkling-Small: 38.4 (#59)

| Benchmark | GPT-5.6 Terra | Inkling-Small |
|---|---|---|
| LMArena Vision | 1271 | 1235 |
| Blueprint-Bench 2 | 30.8% | — |
| Furniture Assembly | 54.2% | — |
| LMArena Document | 1472 | — |

## Multilingual

- GPT-5.6 Terra: 54.5 (#42)
- Inkling-Small: 51.7 (#102)

| Benchmark | GPT-5.6 Terra | Inkling-Small |
|---|---|---|
| LMArena Non-English | 1439 | 1400 |
| LMArena Chinese | 1514 | 1458 |
| LMArena French | 1476 | 1438 |
| LMArena German | 1462 | 1412 |
| LMArena Japanese | 1446 | 1399 |
| LMArena Korean | 1420 | 1355 |
| LMArena Russian | 1450 | 1392 |
| LMArena Spanish | 1455 | 1426 |

## Instruction Following

- GPT-5.6 Terra: 76.6 (#38)
- Inkling-Small: 73.9 (#112)

| Benchmark | GPT-5.6 Terra | Inkling-Small |
|---|---|---|
| LMArena Instruction Following | 1453 | 1397 |

## Long Context

- GPT-5.6 Terra: 44.4 (#64)
- Inkling-Small: 42.7 (#113)

| Benchmark | GPT-5.6 Terra | Inkling-Small |
|---|---|---|
| LMArena Longer Query | 1452 | 1401 |

## Writing & Preference

- GPT-5.6 Terra: 70.4 (#21)
- Inkling-Small: 59.8 (#102)

| Benchmark | GPT-5.6 Terra | Inkling-Small |
|---|---|---|
| LMArena Text | 1446 | 1413 |
| LMArena Creative Writing | 1411 | 1330 |
| EQ-Bench Creative Writing | 1855 | 1491 |
| LMArena Multi-Turn | 1451 | 1420 |
| EQ-Bench 4 | 1234 | — |

## FAQ

### Is GPT-5.6 Terra better than Inkling-Small?

GPT-5.6 Terra is the stronger model overall, scoring 59.1 to 46.5 on the Noometry Index. Inkling-Small costs 7.1× less per token, which makes it the better buy when GPT-5.6 Terra's lead doesn't matter for your workload.

### Which is cheaper, GPT-5.6 Terra or Inkling-Small?

Inkling-Small is cheaper. It lists at $0.45 per million input tokens and $1.20 per million output tokens; GPT-5.6 Terra lists at $2 and $12.

### Is GPT-5.6 Terra or Inkling-Small better for coding?

GPT-5.6 Terra scores higher on coding benchmarks: 57.1 versus 43.3 in the Noometry coding category.

### Which has the bigger context window?

GPT-5.6 Terra does, with 1.05M tokens against 524K.

### How many benchmarks do GPT-5.6 Terra and Inkling-Small share?

33 benchmarks have published results for both models. GPT-5.6 Terra has 52 scored results on Noometry and Inkling-Small has 33.