AI Model Release Tracker

Nemotron 3.5 Bolt 74BOpen Weight

Released

Nemotron 3.5 Bolt 74B is an AI model released by NVIDIA on Thursday, Oct 15 2026, the same day as Nemotron 3.5 Super VL. It is an open-weight model — the trained weights are available to download and run. It is a 74B parameter model. Benchmark results (shown below) cover Terminal-Bench 2.1, MCP Atlas, BrowseComp, and GPQA Diamond.

Benchmarks

Terminal & CLI

Terminal-Bench 2.1Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Higher is better.
50.98%
#35 of 35

Agentic & tool use

MCP AtlasMulti-step tool use — Can the AI chain together many tools and steps to complete one bigger task, rather than doing just a single thing? Higher is better.
52.2%
#13 of 13
BrowseCompWeb browsing — Can the AI browse the web and track down hard-to-find answers? Higher is better.
66.75%
#26 of 30

Reasoning & science

GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better.
86.43%
#28 of 61

Compare Nemotron 3.5 Bolt 74B with

Nemotron 3.5 Bolt 74B

Suggested comparisons

Nemotron 3.5 Bolt 74BvsNemotron 3.5 Super VLNemotron 3.5 Bolt 74BvsGPT-6.1 SolNemotron 3.5 Bolt 74BvsClaude Haiku 5.5Nemotron 3.5 Bolt 74BvsGemini 4 ArgonNemotron 3.5 Bolt 74BvsMuse Spark OpenNemotron 3.5 Bolt 74BvsGrok 4.7Nemotron 3.5 Bolt 74BvsDeepSeek-V4.1-FlashNemotron 3.5 Bolt 74BvsMistral Large 4Nemotron 3.5 Bolt 74BvsKimi K3Nemotron 3.5 Bolt 74BvsGLM-5.3-FlashNemotron 3.5 Bolt 74BvsQwen3.8-Max-0902

About

Nemotron 3.5 Bolt 74B, released October 15, 2026, brought NVIDIA's Super VL architecture to local agents through model compression and quantization. The released 24GB checkpoint used IQ2 weights, with 74 billion total parameters and 9 billion active, while retaining text, image and video inputs. Its benchmark results belong to that quantized checkpoint; the separate BF16 model scored differently.

At launch NVIDIA reported 86.43% on GPQA Diamond, 50.98% on Terminal-Bench 2.1, 66.75% on BrowseComp and 52.20% on MCP Atlas. The MCP Atlas evaluation used the 500 public tasks with reasoning enabled, and BrowseComp used Exa web search. These were NVIDIA-reported evaluations, rather than independent leaderboard measurements.

Frequently asked questions

Nemotron 3.5 Bolt 74B was released by NVIDIA on Thursday, Oct 15 2026.

All NVIDIA releases

11 tracked