Koala
Berkeley's 2023 dialogue-tuned LLaMA — a research-era milestone
Overview
UC Berkeley BAIR's April 2023 LLaMA-13B fine-tune, trained on curated web dialogue such as ShareGPT. Blind testers rated it level with ChatGPT about half the time; it was never commercially licensed and survives only as a research artefact.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 11 Jul 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
Use cases
ASEAN Perspective
Koala in Southeast Asia
ASEAN-region availability and pricing notes coming soon. Drop the editorial team a note via /contact/ if you can supply local context (Singapore/Malaysia/Indonesia/Thailand/Vietnam).
Koala is a 2023 research chatbot from Berkeley's BAIR lab — a LLaMA-13B fine-tune on curated dialogue data that matched ChatGPT in about half of blind comparisons and helped establish that data quality beats data volume for instruction tuning. As a milestone in the open-LLM story it's genuinely significant. As a tool it no longer exists in any practical sense: the demo is offline, the weights were distributed only as diffs against Meta's research-licensed LLaMA, and there's no API, no maintenance, and no commercial path. Current open models — Llama 3.x, Qwen, Gemma — outclass it completely. Read the BAIR blog post for the history; deploy something current.
What people say
April 2023 produced open-source chatbots faster than anyone could evaluate them, and Koala was one of the month's better entries. Berkeley's BAIR lab fine-tuned LLaMA-13B on curated dialogue — ShareGPT conversations, open chat datasets, Q&A — betting that a small amount of high-quality conversational data would beat bulk scraping. The bet paid off in blind tests: users preferred Koala to Stanford's Alpaca, and rated it at least as good as ChatGPT in roughly half of comparisons.
Then the field moved on, mostly to Koala's sibling. Vicuna, released the same week by the adjacent LMSYS group, took the mindshare and the downloads; LMSYS's own chat logs show Vicuna dominating usage while Koala trailed. The Koala demo is long offline, the weights were only ever distributed as diffs against Meta's research-licensed LLaMA — so never commercially usable — and there was no version 2.
Its footprint survives in method rather than product. Koala was early, credible evidence that data curation mattered more than data volume for dialogue tuning, a finding that fed into how later instruction datasets got built. For anyone writing the history of the open-LLM spring of 2023 — Alpaca, Vicuna, Koala, WizardLM — it's a legitimate primary source with a well-written BAIR blog post attached. For anyone building anything: a 13B research artefact three base-model generations behind Llama 3.x, Qwen or Gemma has no practical role. The 3.2 score is honest. This entry exists for the record, not the roadmap.
Summary of public user & expert reviews, compiled by RECATOOLS.
Notable facts
- Koala's training data included real human conversations from real social media and Q&A platforms, giving it exposure to a wider range of conversational styles than synthetic instruction data alone.
- Berkeley researchers found that Koala beat Alpaca on 44% of human evaluation comparisons while losing on only 26%, demonstrating that conversational data quality matters for chatbot applications.
- The naming of Koala, along with Llama, Alpaca, Vicuna, and others, established the unofficial 'zoo' naming convention for open-source language models.
Frequently asked questions
About this listing
This entry was compiled from publicly available data including Koala's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Koala unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Koala directly →
Spotted something out of date? Suggest an update →
Koala in the news
Alternatives to Koala
More in LLMs & Chat
