Introducing Inkling-Small - Thinking Machines Lab
Introducing Inkling-Small
Jul 30, 2026
Try on Tinker
Model card
Hugging Face
Today, we are releasing Inkling-Small, an efficient open-weights model that achieves comparable performance to Inkling at a quarter of its size.
Inkling-Small is a Mixture-of-Experts transformer with 276B total parameters, 12B active, trained on NVIDIA GB300 NVL72 systems. Like Inkling, it features native reasoning over audio and images, variable thinking effort, a context window of up to 1M tokens, and well-rounded performance across a range of benchmarks.
Compared to Inkling, Inkling-Small achieves comparable performance with much less compute. Across agentic tool use (Terminal-Bench 2.1), reasoning (HLE text-only, no tools), and instruction following (IFBench) benchmarks, Inkling-Small is more efficient than Inkling and competitive with other models in its weight class. Furthermore, its variable thinking effort lets users easily adapt it to target their use case, balancing cost and performance.
TFLOPs<br>Cost
Inkling-Small 12B active (effort sweep)<br>Inkling 41B active (effort sweep)<br>Comparison models
Terminal-Bench 2.1
20%<br>40%<br>60%<br>80%
50<br>100<br>200<br>500<br>1k<br>2k<br>5k
Nemotron 3 Super<br>Nemotron 3 Ultra<br>DeepSeek V4 Flash<br>Qwen3.5-397B-A17B<br>MiMo V2.5<br>Minimax M2.7<br>Kimi K2.5<br>Kimi K2.6<br>GLM 5.2
minimal<br>low<br>medium<br>high<br>xhigh
Humanity's Last Exam (Text-Only)
0%<br>10%<br>20%<br>30%<br>40%
20<br>50<br>100<br>200<br>500<br>1k<br>2k<br>5k
Nemotron 3 Super<br>Nemotron 3 Ultra<br>DeepSeek V4 Flash<br>Qwen3.5-397B-A17B<br>MiMo V2.5<br>Minimax M2.7<br>Kimi K2.5<br>Kimi K2.6<br>GLM 5.2<br>gpt-oss-120b
minimal<br>low<br>medium<br>high<br>xhigh
IFBench
60%<br>70%<br>80%<br>90%
20<br>50<br>100<br>200<br>500
Nemotron 3 Super<br>Nemotron 3 Ultra<br>DeepSeek V4 Flash<br>Qwen3.5-397B-A17B<br>MiMo V2.5<br>Minimax M2.7<br>Kimi K2.5<br>Kimi K2.6<br>GLM 5.2<br>gpt-oss-120b
minimal<br>low<br>medium<br>high<br>xhigh
pass@1<br>Approx. Output TFLOPs / task
Inkling-Small 12B active (effort sweep)<br>Inkling 41B active (effort sweep)<br>Comparison models
Terminal-Bench 2.1
20%<br>40%<br>60%<br>80%
0.2c<br>0.5c<br>1c<br>2c<br>5c<br>10c<br>20c<br>50c
Nemotron 3 Super<br>Nemotron 3 Ultra<br>DeepSeek V4 Flash<br>Qwen3.5-397B-A17B<br>MiMo V2.5<br>Minimax M2.7<br>Kimi K2.5<br>Kimi K2.6<br>GLM 5.2
minimal<br>low<br>medium<br>high<br>xhigh
Humanity's Last Exam (Text-Only)
0%<br>10%<br>20%<br>30%<br>40%
0.1c<br>0.2c<br>0.5c<br>1c<br>2c<br>5c<br>10c<br>20c
Nemotron 3 Super<br>Nemotron 3 Ultra<br>DeepSeek V4 Flash<br>Qwen3.5-397B-A17B<br>MiMo V2.5<br>Minimax M2.7<br>Kimi K2.5<br>Kimi K2.6<br>GLM 5.2<br>gpt-oss-120b
minimal<br>low<br>medium<br>high<br>xhigh
IFBench
60%<br>70%<br>80%<br>90%
0.1c<br>0.2c<br>0.5c<br>1c<br>2c<br>5c
Nemotron 3 Super<br>Nemotron 3 Ultra<br>DeepSeek V4 Flash<br>Qwen3.5-397B-A17B<br>MiMo V2.5<br>Minimax M2.7<br>Kimi K2.5<br>Kimi K2.6<br>GLM 5.2<br>gpt-oss-120b
minimal<br>low<br>medium<br>high<br>xhigh
pass@1<br>Estimated output cost ($) / sample
Performance-Compute Comparison. Sweeping reasoning effort from minimal to xhigh traces the performance-compute curve (output TFLOPs per sample) for Inkling-Small and Inkling on Terminal-Bench 2.1, HLE (no tool), and IFBench. We show that Inkling-Small is competitive with other open-weights models in a similar size range on both performance and efficiency. Output TFLOPs per sample are estimated as 2 × active parameters × mean generated tokens per sample, where generated-token counts include reasoning tokens and come from our evaluations or public reports from Artificial Analysis.Performance-Cost Comparison. Sweeping reasoning effort from minimal to xhigh traces the performance-cost curve (dollar output price per sample) for Inkling-Small and Inkling on Terminal-Bench 2.1, HLE (no tool), and IFBench. We show that Inkling-Small is competitive with other open-weights models in a similar size range on both performance and efficiency. Estimated output cost per sample is computed as mean generated tokens per sample × output price per token, where generated-token counts include reasoning tokens and come from our evaluations or public reports from Artificial Analysis (reasoning + answer tokens only). Inkling output pricing is $4.05 / 1M tokens and Inkling-Small output pricing is $1.20 / 1M tokens; for comparison-model pricing, we use the model provider’s official pricing when possible. When this is not possible, we use pricing from the recommended third party inference provider or pricing from Artificial Analysis, whichever is lower.
We are releasing the full weights of Inkling-Small. We’re also making it available for fine-tuning on Tinker, and for text, image, and audio chat on Tinker Playground.
Capabilities
As we build our model family, we are always iterating on our approach. Inkling-Small began training after its larger counterpart, which let us improve its training process. For example, we made changes to Inkling-Small’s pre-training data mix and machine learning recipe. Additionally, we post-trained an earlier checkpoint, Inkling-Small (preview), in part using on-policy distillation with Inkling as the teacher. Starting from that checkpoint, we continued scaling agentic coding RL for two weeks. With these improvements, Inkling-Small surpassed Inkling on reasoning...