Nemotron 120B hits 17 tok/s on AMD Strix Halo

// 66d agoBENCHMARK RESULT

Nemotron 120B hits 17 tok/s on AMD Strix Halo

Early benchmarks show NVIDIA's Nemotron-3 Super 120B-A12B model running at 14-17 tokens per second on AMD's flagship Strix Halo platform. By leveraging 128GB of unified LPDDR5x memory and the Radeon 8060S iGPU, the mobile workstation chip enables massive 384k context windows that previously required multi-GPU server setups.

// ANALYSIS

Strix Halo is proving to be a genuine Apple Silicon competitor for local LLM workloads, finally breaking the 24GB VRAM barrier for PC laptops. Its unified memory architecture allows the iGPU to address nearly the entire 128GB RAM pool, making massive 120B+ models viable on a single mobile chip. Performance of 17 tok/s on a 120B MoE model puts it in direct competition with high-end Mac Studio (M2/M3 Ultra) configurations.

// TAGS

amdryzen-ai-max-plus-395strix-halogpullmlocal-llmopen-weightsnemotron

DISCOVERED

66d ago

2026-03-23

PUBLISHED

66d ago

2026-03-22

RELEVANCE

8/ 10

AUTHOR

Mediocre_Paramedic22

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

NEWS26m ago

Dev lets Claude trade BTC overnight, nets $95 profit

A developer gave Claude a $20 budget to autonomously script and execute Bitcoin trades overnight, waking up to a functional trading bot and a $95 profit across five trades.

OPEN SOURCE1h ago

Plannotator 0.19.24 adds Amp support and configurable storage

Plannotator 0.19.24 is a substantial release that expands the tool beyond Claude Code with native Amp support, adds a `PLANNOTATOR_DATA_DIR` override so users can move the default `~/.plannotator` data directory, introduces Auto Mode in the permission selector for newer Claude Code versions, and fixes a Pi approval crash after plan acceptance. The update folds multiple stacked PRs into one release and pushes the project further toward a multi-agent review layer rather than a single-agent hook utility.

UPDATE2h ago

Grok Build widens access, adds subagents

xAI’s Grok Build is an early-beta terminal coding agent with plan-review-approve flows, parallel subagents, worktree isolation, and support for plugins, hooks, skills, and MCP. The latest improvements make it feel less like a demo and more like xAI’s bid to compete seriously in the AI coding CLI race.