None
None
Long Story Video Skill
Turn your idea into 10s–10min video—AI writes scripts, prompts, and generates footage automatically.
Ads Video Skill
Generate professional ads and sales videos—AI auto-generates scripts, prompts, and footage.
3D Science Explainer Video Skill
Convert scientific concepts into stunning 3D explain animations
Feedback
😢 Every free AI generation costs us real GPU money. We pay out of pocket to keep this free for you.
💎 Upgrade to Pro and unlock:
- ✓HD Download
- ✓No Watermark
- ✓Unlimited Generations
Inkling 2 Small
Inkling 2 Small scores 31 on the Artificial Analysis Intelligence Index and runs in Tinker Playground today, ahead of an Apache 2.0 weight release.
All Tools
Discover our comprehensive AI-powered animation toolkit
Why Use Inkling 2 Small
A second-generation sparse Mixture-of-Experts model with native image input, variable effort, and a 1M-token context window.
- Sparse Mixture-of-Experts at 594B / 17BA 48-layer decoder-only transformer routes every token to 8 of 384 experts, with hybrid local and global attention layers.
- Variable Effort and 1M-Token ContextDial effort per request, feed up to 1M tokens, and send images from 40px to 4096px per side for the best results.
- Closed Weights First, Apache 2.0 NextIt launches on Tinker with closed weights, then full weights follow under Apache 2.0 once external safety testing and red-teaming finish.
How to Use Inkling 2 Small
Three ways to get started with the model on Tinker.
Inkling 2 Small Capabilities
Long-horizon autonomy, vision-in-the-loop design work, and calibrated reasoning carry across coding, research and knowledge tasks.
Long-Horizon Agentic Coding
RL over hundreds of thousands of long-horizon coding tasks across agent harnesses and compaction strategies, graded on approach rather than passing tests alone.
Kernel and ML Autoresearch
Agentic GPU kernel optimization on PMPP-Hard, plus a run that trained Qwen3.5-4B on Tinker for 4.5 hours and $426 to more than double its forecasting score.
Vision Encoder and Visual Grounding
A 292M-parameter Perception Encoder vision stack with 20×20-pixel patches and 2×2 spatial pooling, strong on OCR, grounding, infographics and visual reasoning.
Vision-in-the-Loop Output Validation
It reads its own rendered output to iteratively build websites, games and 3D designs, and named three wines from a single photo in 19 of 20 trials.
Agentic Knowledge Work and Finance
Tool calls over MCP, an ELO of 1527 on AA-Briefcase, and the strongest open-weights results across several finance benchmarks on filings and spreadsheets.
Alignment and Two-Layer Monitoring
CBRN scores below the open-weight frontier, a lower Average Risk Score on FORTRESS, plus a synchronous pre-execution tool monitor and an asynchronous LLM judge.
Inkling 2 Small — FAQ
Common questions and answers about the model.
How large is Inkling 2 Small?
594B total parameters with 17B active per token, routed to 8 of 384 experts inside a 48-layer decoder-only transformer.
When do the open weights arrive?
Weights stay closed on Tinker at launch; full weights are intended for an Apache 2.0 release the following week, after external safety testing and red-teaming.
What inputs and context does it support?
UTF-8 text and pixel-based images, ideally 40px to 4096px per dimension, with text output and a context window of up to 1M tokens.
How was the model trained?
8.4T pretraining tokens in 68 hours on NVIDIA GB300 NVL72 racks in native NVFP4, a 3.8T-token phase at 512K context, 320B reasoning and agent tokens at 1M context, 34B top-quality agent trajectories, then a joint RL run of 1,075 steps and over 17 million rollouts in 9 days.
Which benchmarks does it lead on?
A score of 31 on the Artificial Analysis Intelligence Index, ELO 1527 on AA-Briefcase, best-among-open-weights finance results, and a pending top DesignArena Website score.
Can it be customized to my data?
Yes — it learns faster than earlier models on the same data and recipe, matching Inkling Small's final score on a financial-document summarization task in just 75 training steps.
Try Inkling 2 Small in Tinker Playground Today
Open the Tinker Playground to run the model now, start fine-tuning with the Tinker Cookbook, and watch for the Apache 2.0 weight release.