Inkling
Inkling is an open-weights language model from Thinking Machines, released July 15, 2026. It has 952B parameters (mixture-of-experts, 18B active). On the Epoch Capabilities Index it scores 148.4, rank 76 of 256 rated models.
Our mission is to build AI that extends human will and judgment. We have developed a platform that lets anyone customize models, previewed an AI system built for interactive collaboration, and published novel research.
Check it against your own PC →
Can you run Inkling locally?
Even with 32 GB of VRAM and 32 GB of RAM it is too large to run at a usable speed on a single consumer GPU. Estimates assume 32 GB of DDR5 RAM, a 16K context and llama.cpp-style GGUF quantization. Enter your exact hardware in the live app.
| GPU memory | Verdict | Best quant | VRAM | RAM | tok/s |
|---|---|---|---|---|---|
| 6 GB VRAM | Too large | – | – | – | – |
| 8 GB VRAM | Too large | – | – | – | – |
| 12 GB VRAM | Too large | – | – | – | – |
| 16 GB VRAM | Too large | – | – | – | – |
| 24 GB VRAM | Too large | – | – | – | – |
| 32 GB VRAM | Too large | – | – | – | – |
Ready-made quantizations on Hugging Face: unsloth/inkling-GGUF · thinkingmachines/Inkling-NVFP4 · RedHatAI/Inkling-NVFP4-FP8-BLOCK · thunquant/Inkling-NVFP4 · schwyzquants/Inkling-NVFP4 · bielquants/Inkling-NVFP4
Strengths by category
ModelTape's 0–100 fit on Epoch AI benchmark results.
| Category | Score | Based on |
|---|---|---|
| Reasoning | 13.9 | Chess Puzzles, SimpleBench |
| Math | 31.2 | ProofBench, OTIS Mock AIME 2024-2025, FrontierMath-Tiers-1-3-v2-Private, FrontierMath-Tier-4-v2-Private |
| Knowledge | 71.9 | GPQA diamond, SimpleQA Verified |
Benchmark results
| Benchmark (Epoch AI) | Score |
|---|---|
| OTIS Mock AIME | 88.9% |
| GPQA Diamond | 87.9% |
| SimpleQA Verified | 40.3% |
| SimpleBench | 40.0% |
| FrontierMath (Tiers 1–3) | 35.1% |
| Chess Puzzles | 16.9% |
| FrontierMath-Tier-4-v2-Private | 4.9% |
| FrontierSWE | 4.1% |
| ProofBench | 0.0% |
Links
Announcement · Hugging Face: thinkingmachines/Inkling · Epoch AI benchmarks
Models with a similar score
Gemini 3 Flash · Grok 4.20 · Claude Opus 4.5 · Kimi K2.6 · GPT-5 Pro · Kimi K2.7 Code · GPT-5.1 · Beam 1.0