Particle-1

Particle's coding model post-trained from DeepSeek V4.1 Flash to cut reasoning and tool-call tokens in agentic software work.

Particle · Developer tools and coding agents · model

Published in Feed

Overview

Particle-1 is a software-engineering model from Particle, released October 28, 2026, post-trained from DeepSeek V4.1 Flash using reinforcement-learning environments Particle built. OpenRouter describes it as a token-efficient agentic coding model aimed at terminal workflows, codebase changes and data engineering, designed to use fewer reasoning and tool-call tokens while keeping accuracy near the base model. OpenRouter lists a 1.05M-token context window at $0.25 per million input tokens and $1 per million output tokens. Efficiency and task-performance claims are Particle's own; no independent benchmark of Particle-1 was found.

This is a sourced directory entry. Features and capabilities are reported in the linked sources; this entry does not document independent Aymorphic testing.

Key features

  • Post-trained from DeepSeek V4.1 Flash with Particle's RL environments
  • Targets fewer reasoning and tool-call tokens at near-base accuracy (company claim)
  • 1.05M-token context window on OpenRouter
  • Initial release focused on software engineering: terminal, codebase edits, data engineering

Use cases

  • Cost-sensitive autonomous coding agents running long tool-call loops
  • Terminal and repository-maintenance workflows
  • Data-engineering task automation

Pricing

OpenRouter: $0.25 per million input tokens, $1 per million output tokens.

Pricing reflects the source record and may change. Confirm current terms with the provider.

Dates and sources

Released
28 Oct 2026
Record updated
2 Nov 2026
Record last checked
2 Nov 2026

Explore the Feed

Other entries filed under developer tools and coding agents; inclusion does not imply equivalent features.

  • Claude Skills Marketplace

    Anthropic's public marketplace for Claude Agent Skills, letting developers publish, monetize, and install reviewed Skills directly into Claude Code and the Claude apps.

  • Composer 2

    Cursor's second-generation in-house coding model, now the default agent model in the editor, with faster multi-file edits and native terminal tool use.

  • Devin 3

    Cognition's autonomous software engineering agent, now able to run unsupervised multi-day engineering sprints with persistent repo memory and automatic PR-review handling.

Browse the Feed