Inkling
Inkling (Thinking Machines) was released on Jul 15, 2026. In the Epoch AI capability ranking (ECI), it ranks #66 out of 267 models, with 148.59. On Arena, which measures human preference in blind comparisons, it has 1,492 ELO points (#66). On LiveBench, the overall average is 71.9, #44 out of 58 scored models.
Where Inkling stands out and where it falls behind, by the numbers
Inkling has published results on 20 tests compiled by Epoch AI. The yardstick here is its generation: its position among the models released between Oct 18, 2025 and Oct 13, 2026 that took the same test. Against the whole archive, which includes 2023 models, a recent release lands near the top of almost everything, and that says nothing.
Where it stands out
Tests where it sits in the top third of its own generation.
It is not in the top third of its generation on any test with 6 or more models compared.
Where it falls behind
Tests where it sits in the bottom half of its own generation.
See all 20 tests, one by one
| Test | Score | In its generation | Overall | Test record |
|---|---|---|---|---|
| GPQA Diamond | 88.3% | #50 / 80 | #51 / 173 | 96.4% Gemini 4 Argon |
| AIME 2024 and 2025 (OTIS) | 88.9% | #57 / 82 (tied with 2) | #59 / 168 | 100.0% GPT-5.5 |
| Dtbench | 87.5% | #48 / 69 (tied with 1) | #50 / 148 | 98.9% Claude Opus 5.5 |
| Lmca | 37.58 | #51 / 69 (tied with 1) | #55 / 121 | 68.23 Claude Opus 5.5 |
| CritPt | 5.4% | #58 / 79 (tied with 1) | #59 / 112 | 32.3% GPT-5.6 Sol |
| WeirdML | 32.3% | #54 / 54 | #88 / 112 | 93.6% GPT-6 Astra |
| SciCode | 47.0% | #57 / 76 | #57 / 104 | 66.9% Claude Opus 5.5 |
| Frontiermath Tiers 1 3 V2 | 33.3% | #59 / 67 | #64 / 86 | 93.7% GPT-6 Astra |
| ARC-AGI 2 | 36.5% | #39 / 50 | #39 / 83 | 95.0% GPT-6 Astra |
| ALE-Bench | 946 | #41 / 64 | #42 / 82 | 2,951 GPT-6 Astra |
| WebDev Arena | 1,410 | #57 / 75 | #58 / 82 | 1,820 Claude Opus 5.5 |
| ARC-AGI | 79.5% | #40 / 50 | #40 / 81 | 98.5% Claude Fable 5 |
| SimpleQA Verified | 40.3% | #42 / 63 | #48 / 81 | 76.0% Gemini 4 Argon |
| Chess puzzles | 21.0% | #49 / 64 (tied with 4) | #55 / 80 | 76.0% GPT-6 Astra |
| ProofBench | 0.0% | #70 / 70 | #74 / 74 | 100.0% Claude Fable 5.1 |
| SimpleBench | 50.0% | #29 / 38 | #37 / 74 | 81.9% Claude Fable 5 |
| Frontiermath Tier 4 V2 | 4.9% | #53 / 57 (tied with 1) | #57 / 67 | 100.0% GPT-6.1 Sol |
| Apex Agents | 33.8% | #43 / 49 | #43 / 50 | 82.4% Gemini 4 Argon |
| FrontierCode | 14.0% | #40 / 43 | #40 / 43 | 54.6% Claude Opus 5.5 |
| FrontierSWE | 4.1% | #20 / 20 | #20 / 20 | 65.5% GPT-6 Astra |
For each test, the model's best effort level counts, as in the ranking. Source of the scores: Epoch AI, under CC BY 4.0.
What to compare Inkling with
The models from other labs closest in capability and released in the same period, at most two per lab. Click any of them to see its page.
| Model | Lab | ECI | ELO | LiveBench | Release |
|---|---|---|---|---|---|
| Qwen 3.6 Max (Preview) | Alibaba | 149.24 | 1,488 | n/a | Apr 20, 2026 |
| Grok 4.3 (no thinking) | xAI | 149.21 | 1,425 | 62.3 | Apr 17, 2026 |
| DeepSeek-V4-Pro | DeepSeek | 149.19 | 1,486 | 71.6 | Apr 24, 2026 |
| Grok 4.3 Beta | xAI | 148.95 | n/a | n/a | Apr 17, 2026 |
| GPT-5.4 Mini | OpenAI | 148.77 | 1,446 | 66.4 | Mar 17, 2026 |
| Inkling | Thinking Machines | 148.59 | 1,492 | 71.9 | Jul 15, 2026 |
| Kimi K2.5 | Moonshot | 147.98 | 1,495 | n/a | Jan 27, 2026 |
Inkling or GPT-5.4 Mini?
GPT-5.4 Mini is the model from another lab closest to Inkling in capability. Across the 17 tests both took, this is the score:
| Test | Inkling | GPT-5.4 Mini |
|---|---|---|
| ProofBench | 0.0% | 21.0% |
| Frontiermath Tier 4 V2 | 4.9% | 9.8% |
| ARC-AGI 2 | 36.5% | 18.9% |
| FrontierCode | 14.0% | 27.0% |
| WeirdML | 32.3% | 60.3% |
| CritPt | 5.4% | 10.0% |
| Frontiermath Tiers 1 3 V2 | 33.3% | 51.2% |
| SimpleQA Verified | 40.3% | 29.4% |
| ALE-Bench | 946 | 1,189 |
| ARC-AGI | 79.5% | 63.7% |
Unweighted score: each test counts one point, whether math or coding. The dot marks who did better on each row.
Other models from Thinking Machines
The models from the same lab closest in time, newest first.
| Model | Release | ECI | ELO | LiveBench | Access |
|---|---|---|---|---|---|
| Inkling-Small | Jul 15, 2026 | 150.18 | 1,463 | n/a | Open weights (unrestricted) |
| Inkling | Jul 15, 2026 | 148.59 | 1,492 | 71.9 | Open weights (unrestricted) |
Have you used Inkling? Rate it
Users' opinion, not measurement: the rating is kept apart from the ranking and does not move anyone's position. One vote per browser, no sign-up, and you can change your rating later.
What people ask about Inkling
Where does Inkling rank among AI models?
In the ROO3 AI Benchmark, ordered by the Epoch AI ECI, Inkling ranks #66 out of 267 models, with an ECI of 148.59. The ECI is a composite index that combines several hard benchmarks into a single scale. The position changes when a new model comes out, and the page is refreshed six times a day.
Is Inkling better than GPT-5.4 Mini?
Across the 17 tests both models took and Epoch AI compiled, Inkling comes out ahead on 6, behind on 10 and tied on 1. It is a test-by-test count with no weighting: a win in math counts the same as a win in coding. To choose between the two, look at the tests that resemble your own use.
Where do the numbers on this page come from?
From the same three sources as the ranking, all openly licensed: Epoch AI (ECI and individual tests, CC BY 4.0), Arena (human-preference ELO, CC BY 4.0) and LiveBench (overall average, CC BY-SA 4.0). ROO3 does not measure any model. When the page shows a figure published by the maker itself, it appears in the written analysis, labeled as the maker's, and never enters the ranking.
Where each number comes from
Epoch AI, "AI Benchmarking Hub". Published at epoch.ai. Data under the CC BY 4.0 licence, reordered by roo3.co.
The Arena (LMArena) leaderboard, from the public lmarena-ai/leaderboard-dataset on Hugging Face, under the CC BY 4.0 licence. We are not affiliated with the Arena.
LiveBench, by the LiveBench team, published at livebench.ai under the CC BY-SA 4.0 license. The overall average is recomputed by roo3.co with the same formula as the website (average of the categories), and those numbers stay under CC BY-SA 4.0. We are not affiliated with LiveBench.
← See the full ranking