nowfound

AI · October 5, 2026

I finetuned 1.5B Qwen to near GPT-4o level bash generation perf

EasyCommand: open models, training data, and lessons from teaching small models to write shell commands.

I finetuned 1.5B Qwen to near GPT-4o level bash generation perf launched on October 5, 2026 with 6 votes, #733 of 3,603 launches that month and more than 75% of that year's launches.

Visit dirac.run ↗Alternativestop 20% of October 2026

What it does

Tl;Dr: Fine-tuned Qwen2.5-Coder-1.5B-Instruct and Qwen3-0.6B into CPU-friendly English-to-Bash models. In local development tests, the released 1.5B 4-bit model passed 212/300 ALFA-updated tasks (70.7%), versus whatisit's 191/300 (63.7%): a 7-percentage-point gap. Both used the same grading definition, but different prompts and generation settings. The comparison table includes both original and updated ALFA scores, with the benchmark version marked. Despite using LLMs for most of the coding, there was always one thing I kept Googling: Bash commands. It's quite flow-breaking to pause work, open Google, type the full query, go to Stack Overflow or similar, and look up the syntax I wanted.…from dirac.run

Does a similar job

all alternatives →
  • Qwen3.5 SmallMar 2026 · huggingface.co · ▲355

    0.8B-9B native multimodal w/ more intelligence, less compute

  • BashBuddy2025 · bashbuddy.run · ▲159

    Write bash commands with natural language, fully local.

  • Qwen3.6-35B-A3BApr 2026 · qwen.ai · ▲124

    The open sparse MoE model for agentic coding

  • Qwen3.8-MaxAug 2026 · qwen.ai · ▲277

    Qwen3.8-Max + MCP for coding on your local machine, without paying for Qwen Code. Qwen3.8-Max itself runs in the cloud through Qwen Studio — this setup just gives it access to your local files and terminal through MCP. My take: it is slower than Codex and Claude Code, especially in thinking mode, but it does work well. MCP adds some latency too since there’s an extra layer between the model and your machine. In fast mode, it gets noticeably closer. I wouldn’t put it in the same category as Fable or Opus-5. Personally, I’d place it closer to the next Opus 4.7 level in terms of coding ability.…

  • Qwen3-Coder2025 · qwenlm.github.io · ▲248

    A powerful open model for agentic coding tasks

  • GPT-3 Powered Shell2022 · musings.yasyf.com · ▲308

    Use gpt-do to have GPT-3 write all your bash commands for you.

More ai this month

the category →
  • Spira Maxima▲578

    Video model that turns script into viral social media videos

    AI · 19d ago · spira.ai

  • Rill Browser▲540

    The browser where Claude Code and Codex work beside you

    AI · 19d ago · rill.love

  • Germany's new sovereign AI model Kolibri▲424

    Kolibri is Aleph Alpha

    AI · 21d ago · tej.as

  • Giving Opus 5.5 a simulated paint canvas▲392

    Paintings by AI models: programs run through a simulation of oil paint.

    AI · 22d ago · stillwet.art

  • Beam.win▲379

    Run AI agents on one Mac and use its windows on another

    AI · 5d ago · beam.win

  • FastRouter.ai▲364

    Route requests to the right LLM for cost, latency & quality

    AI · 23d ago · fastrouter.ai

Launched alongside, October 2026

the whole month →